Skip to content

Instantly share code, notes, and snippets.

@pashu123
Created January 12, 2023 17:17
Show Gist options
  • Save pashu123/e1e97043c3deda71716cd8aeb367c837 to your computer and use it in GitHub Desktop.
Save pashu123/e1e97043c3deda71716cd8aeb367c837 to your computer and use it in GitHub Desktop.
graph():
%arg0_1 : [#users=1] = placeholder[target=arg0_1]
%arg1_1 : [#users=1] = placeholder[target=arg1_1]
%arg2_1 : [#users=52] = placeholder[target=arg2_1]
%arg3_1 : [#users=1] = placeholder[target=arg3_1]
%expand : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%arg1_1, [2]), kwargs = {})
%arange : [#users=1] = call_function[target=torch.ops.aten.arange](args = (0, 128), kwargs = {dtype: torch.float32, device: cuda:0, pin_memory: False})
%mul : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%arange, -9.210340371976184), kwargs = {})
%div : [#users=1] = call_function[target=torch.ops.aten.div](args = (%mul, 128), kwargs = {})
%exp : [#users=1] = call_function[target=torch.ops.aten.exp](args = (%div,), kwargs = {})
%slice_1 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%expand, 0, 0, 9223372036854775807), kwargs = {})
%unsqueeze : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_1, 1), kwargs = {})
%unsqueeze_1 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%exp, 0), kwargs = {})
%slice_2 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%unsqueeze_1, 1, 0, 9223372036854775807), kwargs = {})
%mul_1 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%unsqueeze, %slice_2), kwargs = {})
%mul_2 : [#users=2] = call_function[target=torch.ops.aten.mul](args = (%mul_1, 1), kwargs = {})
%sin : [#users=1] = call_function[target=torch.ops.aten.sin](args = (%mul_2,), kwargs = {})
%cos : [#users=1] = call_function[target=torch.ops.aten.cos](args = (%mul_2,), kwargs = {})
%cat : [#users=2] = call_function[target=torch.ops.aten.cat](args = ([%sin, %cos], -1), kwargs = {})
%slice_3 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%cat, 0, 0, 9223372036854775807), kwargs = {})
%slice_4 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_3, 1, 128, 9223372036854775807), kwargs = {})
%slice_5 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%cat, 0, 0, 9223372036854775807), kwargs = {})
%slice_6 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_5, 1, 0, 128), kwargs = {})
%cat_1 : [#users=1] = call_function[target=torch.ops.aten.cat](args = ([%slice_4, %slice_6], -1), kwargs = {})
%_to_copy : [#users=1] = call_function[target=torch.ops.aten._to_copy](args = (%cat_1,), kwargs = {dtype: torch.float16})
%_param_constant0 : [#users=1] = get_attr[target=_param_constant0]
%t : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant0,), kwargs = {})
%_param_constant1 : [#users=1] = get_attr[target=_param_constant1]
%addmm : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant1, %_to_copy, %t), kwargs = {})
%silu : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%addmm,), kwargs = {})
%_param_constant2 : [#users=1] = get_attr[target=_param_constant2]
%t_1 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant2,), kwargs = {})
%_param_constant3 : [#users=1] = get_attr[target=_param_constant3]
%addmm_1 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant3, %silu, %t_1), kwargs = {})
%_param_constant4 : [#users=1] = get_attr[target=_param_constant4]
%embedding : [#users=1] = call_function[target=torch.ops.aten.embedding](args = (%_param_constant4, %arg3_1), kwargs = {})
%add : [#users=22] = call_function[target=torch.ops.aten.add](args = (%addmm_1, %embedding), kwargs = {})
%_param_constant5 : [#users=1] = get_attr[target=_param_constant5]
%_param_constant6 : [#users=1] = get_attr[target=_param_constant6]
%convolution : [#users=3] = call_function[target=torch.ops.aten.convolution](args = (%arg0_1, %_param_constant5, %_param_constant6, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%view : [#users=2] = call_function[target=torch.ops.aten.view](args = (%convolution, [2, 32, 8, 16384]), kwargs = {})
%convert_element_type : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view, torch.float32), kwargs = {})
%var_mean : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem : [#users=1] = call_function[target=operator.getitem](args = (%var_mean, 0), kwargs = {})
%getitem_1 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean, 1), kwargs = {})
%add_1 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem, 1e-05), kwargs = {})
%rsqrt : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_1,), kwargs = {})
%sub : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view, %getitem_1), kwargs = {})
%mul_3 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub, %rsqrt), kwargs = {})
%view_1 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_3, [2, 256, 128, 128]), kwargs = {})
%_param_constant7 : [#users=1] = get_attr[target=_param_constant7]
%unsqueeze_2 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant7, 0), kwargs = {})
%unsqueeze_3 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_2, 2), kwargs = {})
%unsqueeze_4 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_3, 3), kwargs = {})
%_param_constant8 : [#users=1] = get_attr[target=_param_constant8]
%unsqueeze_5 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant8, 0), kwargs = {})
%unsqueeze_6 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_5, 2), kwargs = {})
%unsqueeze_7 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_6, 3), kwargs = {})
%mul_4 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_1, %unsqueeze_7), kwargs = {})
%add_2 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_4, %unsqueeze_4), kwargs = {})
%convert_element_type_1 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_2, torch.float16), kwargs = {})
%convert_element_type_2 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_1, torch.float16), kwargs = {})
%convert_element_type_3 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt, torch.float16), kwargs = {})
%squeeze : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_2, 3), kwargs = {})
%squeeze_1 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze, 2), kwargs = {})
%squeeze_2 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_3, 3), kwargs = {})
%squeeze_3 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_2, 2), kwargs = {})
%detach : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_1,), kwargs = {})
%detach_1 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_3,), kwargs = {})
%silu_1 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_1,), kwargs = {})
%_param_constant9 : [#users=1] = get_attr[target=_param_constant9]
%_param_constant10 : [#users=1] = get_attr[target=_param_constant10]
%convolution_1 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_1, %_param_constant9, %_param_constant10, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_2 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant11 : [#users=1] = get_attr[target=_param_constant11]
%t_2 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant11,), kwargs = {})
%_param_constant12 : [#users=1] = get_attr[target=_param_constant12]
%addmm_2 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant12, %silu_2, %t_2), kwargs = {})
%slice_7 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_2, 0, 0, 9223372036854775807), kwargs = {})
%slice_8 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_7, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_8 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_8, 2), kwargs = {})
%unsqueeze_9 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_8, 3), kwargs = {})
%add_3 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_1, %unsqueeze_9), kwargs = {})
%view_2 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_3, [2, 32, 8, 16384]), kwargs = {})
%convert_element_type_4 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_2, torch.float32), kwargs = {})
%var_mean_1 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_4, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_2 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_1, 0), kwargs = {})
%getitem_3 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_1, 1), kwargs = {})
%add_4 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_2, 1e-05), kwargs = {})
%rsqrt_1 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_4,), kwargs = {})
%sub_1 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_2, %getitem_3), kwargs = {})
%mul_5 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_1, %rsqrt_1), kwargs = {})
%view_3 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_5, [2, 256, 128, 128]), kwargs = {})
%_param_constant13 : [#users=1] = get_attr[target=_param_constant13]
%unsqueeze_10 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant13, 0), kwargs = {})
%unsqueeze_11 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_10, 2), kwargs = {})
%unsqueeze_12 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_11, 3), kwargs = {})
%_param_constant14 : [#users=1] = get_attr[target=_param_constant14]
%unsqueeze_13 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant14, 0), kwargs = {})
%unsqueeze_14 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_13, 2), kwargs = {})
%unsqueeze_15 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_14, 3), kwargs = {})
%mul_6 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_3, %unsqueeze_15), kwargs = {})
%add_5 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_6, %unsqueeze_12), kwargs = {})
%convert_element_type_5 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_5, torch.float16), kwargs = {})
%convert_element_type_6 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_3, torch.float16), kwargs = {})
%convert_element_type_7 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_1, torch.float16), kwargs = {})
%squeeze_4 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_6, 3), kwargs = {})
%squeeze_5 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_4, 2), kwargs = {})
%squeeze_6 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_7, 3), kwargs = {})
%squeeze_7 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_6, 2), kwargs = {})
%detach_2 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_5,), kwargs = {})
%detach_3 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_7,), kwargs = {})
%silu_3 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_5,), kwargs = {})
%_param_constant15 : [#users=1] = get_attr[target=_param_constant15]
%_param_constant16 : [#users=1] = get_attr[target=_param_constant16]
%convolution_2 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_3, %_param_constant15, %_param_constant16, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%add_6 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution, %convolution_2), kwargs = {})
%div_1 : [#users=3] = call_function[target=torch.ops.aten.div](args = (%add_6, 1.0), kwargs = {})
%view_4 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%div_1, [2, 32, 8, 16384]), kwargs = {})
%convert_element_type_8 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_4, torch.float32), kwargs = {})
%var_mean_2 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_8, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_4 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_2, 0), kwargs = {})
%getitem_5 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_2, 1), kwargs = {})
%add_7 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_4, 1e-05), kwargs = {})
%rsqrt_2 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_7,), kwargs = {})
%sub_2 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_4, %getitem_5), kwargs = {})
%mul_7 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_2, %rsqrt_2), kwargs = {})
%view_5 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_7, [2, 256, 128, 128]), kwargs = {})
%_param_constant17 : [#users=1] = get_attr[target=_param_constant17]
%unsqueeze_16 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant17, 0), kwargs = {})
%unsqueeze_17 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_16, 2), kwargs = {})
%unsqueeze_18 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_17, 3), kwargs = {})
%_param_constant18 : [#users=1] = get_attr[target=_param_constant18]
%unsqueeze_19 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant18, 0), kwargs = {})
%unsqueeze_20 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_19, 2), kwargs = {})
%unsqueeze_21 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_20, 3), kwargs = {})
%mul_8 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_5, %unsqueeze_21), kwargs = {})
%add_8 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_8, %unsqueeze_18), kwargs = {})
%convert_element_type_9 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_8, torch.float16), kwargs = {})
%convert_element_type_10 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_5, torch.float16), kwargs = {})
%convert_element_type_11 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_2, torch.float16), kwargs = {})
%squeeze_8 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_10, 3), kwargs = {})
%squeeze_9 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_8, 2), kwargs = {})
%squeeze_10 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_11, 3), kwargs = {})
%squeeze_11 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_10, 2), kwargs = {})
%detach_4 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_9,), kwargs = {})
%detach_5 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_11,), kwargs = {})
%silu_4 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_9,), kwargs = {})
%_param_constant19 : [#users=1] = get_attr[target=_param_constant19]
%_param_constant20 : [#users=1] = get_attr[target=_param_constant20]
%convolution_3 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_4, %_param_constant19, %_param_constant20, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_5 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant21 : [#users=1] = get_attr[target=_param_constant21]
%t_3 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant21,), kwargs = {})
%_param_constant22 : [#users=1] = get_attr[target=_param_constant22]
%addmm_3 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant22, %silu_5, %t_3), kwargs = {})
%slice_9 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_3, 0, 0, 9223372036854775807), kwargs = {})
%slice_10 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_9, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_22 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_10, 2), kwargs = {})
%unsqueeze_23 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_22, 3), kwargs = {})
%add_9 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_3, %unsqueeze_23), kwargs = {})
%view_6 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_9, [2, 32, 8, 16384]), kwargs = {})
%convert_element_type_12 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_6, torch.float32), kwargs = {})
%var_mean_3 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_12, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_6 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_3, 0), kwargs = {})
%getitem_7 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_3, 1), kwargs = {})
%add_10 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_6, 1e-05), kwargs = {})
%rsqrt_3 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_10,), kwargs = {})
%sub_3 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_6, %getitem_7), kwargs = {})
%mul_9 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_3, %rsqrt_3), kwargs = {})
%view_7 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_9, [2, 256, 128, 128]), kwargs = {})
%_param_constant23 : [#users=1] = get_attr[target=_param_constant23]
%unsqueeze_24 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant23, 0), kwargs = {})
%unsqueeze_25 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_24, 2), kwargs = {})
%unsqueeze_26 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_25, 3), kwargs = {})
%_param_constant24 : [#users=1] = get_attr[target=_param_constant24]
%unsqueeze_27 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant24, 0), kwargs = {})
%unsqueeze_28 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_27, 2), kwargs = {})
%unsqueeze_29 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_28, 3), kwargs = {})
%mul_10 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_7, %unsqueeze_29), kwargs = {})
%add_11 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_10, %unsqueeze_26), kwargs = {})
%convert_element_type_13 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_11, torch.float16), kwargs = {})
%convert_element_type_14 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_7, torch.float16), kwargs = {})
%convert_element_type_15 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_3, torch.float16), kwargs = {})
%squeeze_12 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_14, 3), kwargs = {})
%squeeze_13 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_12, 2), kwargs = {})
%squeeze_14 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_15, 3), kwargs = {})
%squeeze_15 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_14, 2), kwargs = {})
%detach_6 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_13,), kwargs = {})
%detach_7 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_15,), kwargs = {})
%silu_6 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_13,), kwargs = {})
%_param_constant25 : [#users=1] = get_attr[target=_param_constant25]
%_param_constant26 : [#users=1] = get_attr[target=_param_constant26]
%convolution_4 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_6, %_param_constant25, %_param_constant26, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%add_12 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%div_1, %convolution_4), kwargs = {})
%div_2 : [#users=2] = call_function[target=torch.ops.aten.div](args = (%add_12, 1.0), kwargs = {})
%_param_constant27 : [#users=1] = get_attr[target=_param_constant27]
%_param_constant28 : [#users=1] = get_attr[target=_param_constant28]
%convolution_5 : [#users=3] = call_function[target=torch.ops.aten.convolution](args = (%div_2, %_param_constant27, %_param_constant28, [2, 2], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%view_8 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%convolution_5, [2, 32, 8, 4096]), kwargs = {})
%convert_element_type_16 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_8, torch.float32), kwargs = {})
%var_mean_4 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_16, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_8 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_4, 0), kwargs = {})
%getitem_9 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_4, 1), kwargs = {})
%add_13 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_8, 1e-05), kwargs = {})
%rsqrt_4 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_13,), kwargs = {})
%sub_4 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_8, %getitem_9), kwargs = {})
%mul_11 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_4, %rsqrt_4), kwargs = {})
%view_9 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_11, [2, 256, 64, 64]), kwargs = {})
%_param_constant29 : [#users=1] = get_attr[target=_param_constant29]
%unsqueeze_30 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant29, 0), kwargs = {})
%unsqueeze_31 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_30, 2), kwargs = {})
%unsqueeze_32 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_31, 3), kwargs = {})
%_param_constant30 : [#users=1] = get_attr[target=_param_constant30]
%unsqueeze_33 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant30, 0), kwargs = {})
%unsqueeze_34 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_33, 2), kwargs = {})
%unsqueeze_35 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_34, 3), kwargs = {})
%mul_12 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_9, %unsqueeze_35), kwargs = {})
%add_14 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_12, %unsqueeze_32), kwargs = {})
%convert_element_type_17 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_14, torch.float16), kwargs = {})
%convert_element_type_18 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_9, torch.float16), kwargs = {})
%convert_element_type_19 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_4, torch.float16), kwargs = {})
%squeeze_16 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_18, 3), kwargs = {})
%squeeze_17 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_16, 2), kwargs = {})
%squeeze_18 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_19, 3), kwargs = {})
%squeeze_19 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_18, 2), kwargs = {})
%detach_8 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_17,), kwargs = {})
%detach_9 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_19,), kwargs = {})
%silu_7 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_17,), kwargs = {})
%_param_constant31 : [#users=1] = get_attr[target=_param_constant31]
%_param_constant32 : [#users=1] = get_attr[target=_param_constant32]
%convolution_6 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_7, %_param_constant31, %_param_constant32, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_8 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant33 : [#users=1] = get_attr[target=_param_constant33]
%t_4 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant33,), kwargs = {})
%_param_constant34 : [#users=1] = get_attr[target=_param_constant34]
%addmm_4 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant34, %silu_8, %t_4), kwargs = {})
%slice_11 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_4, 0, 0, 9223372036854775807), kwargs = {})
%slice_12 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_11, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_36 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_12, 2), kwargs = {})
%unsqueeze_37 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_36, 3), kwargs = {})
%add_15 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_6, %unsqueeze_37), kwargs = {})
%view_10 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_15, [2, 32, 16, 4096]), kwargs = {})
%convert_element_type_20 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_10, torch.float32), kwargs = {})
%var_mean_5 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_20, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_10 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_5, 0), kwargs = {})
%getitem_11 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_5, 1), kwargs = {})
%add_16 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_10, 1e-05), kwargs = {})
%rsqrt_5 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_16,), kwargs = {})
%sub_5 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_10, %getitem_11), kwargs = {})
%mul_13 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_5, %rsqrt_5), kwargs = {})
%view_11 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_13, [2, 512, 64, 64]), kwargs = {})
%_param_constant35 : [#users=1] = get_attr[target=_param_constant35]
%unsqueeze_38 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant35, 0), kwargs = {})
%unsqueeze_39 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_38, 2), kwargs = {})
%unsqueeze_40 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_39, 3), kwargs = {})
%_param_constant36 : [#users=1] = get_attr[target=_param_constant36]
%unsqueeze_41 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant36, 0), kwargs = {})
%unsqueeze_42 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_41, 2), kwargs = {})
%unsqueeze_43 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_42, 3), kwargs = {})
%mul_14 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_11, %unsqueeze_43), kwargs = {})
%add_17 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_14, %unsqueeze_40), kwargs = {})
%convert_element_type_21 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_17, torch.float16), kwargs = {})
%convert_element_type_22 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_11, torch.float16), kwargs = {})
%convert_element_type_23 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_5, torch.float16), kwargs = {})
%squeeze_20 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_22, 3), kwargs = {})
%squeeze_21 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_20, 2), kwargs = {})
%squeeze_22 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_23, 3), kwargs = {})
%squeeze_23 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_22, 2), kwargs = {})
%detach_10 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_21,), kwargs = {})
%detach_11 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_23,), kwargs = {})
%silu_9 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_21,), kwargs = {})
%_param_constant37 : [#users=1] = get_attr[target=_param_constant37]
%_param_constant38 : [#users=1] = get_attr[target=_param_constant38]
%convolution_7 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_9, %_param_constant37, %_param_constant38, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%_param_constant39 : [#users=1] = get_attr[target=_param_constant39]
%_param_constant40 : [#users=1] = get_attr[target=_param_constant40]
%convolution_8 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%convolution_5, %_param_constant39, %_param_constant40, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
%add_18 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_8, %convolution_7), kwargs = {})
%div_3 : [#users=2] = call_function[target=torch.ops.aten.div](args = (%add_18, 1.0), kwargs = {})
%view_12 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%div_3, [2, 32, 16, 4096]), kwargs = {})
%convert_element_type_24 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_12, torch.float32), kwargs = {})
%var_mean_6 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_24, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_12 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_6, 0), kwargs = {})
%getitem_13 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_6, 1), kwargs = {})
%add_19 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_12, 1e-06), kwargs = {})
%rsqrt_6 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_19,), kwargs = {})
%sub_6 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_12, %getitem_13), kwargs = {})
%mul_15 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_6, %rsqrt_6), kwargs = {})
%view_13 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_15, [2, 512, 64, 64]), kwargs = {})
%_param_constant41 : [#users=1] = get_attr[target=_param_constant41]
%unsqueeze_44 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant41, 0), kwargs = {})
%unsqueeze_45 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_44, 2), kwargs = {})
%unsqueeze_46 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_45, 3), kwargs = {})
%_param_constant42 : [#users=1] = get_attr[target=_param_constant42]
%unsqueeze_47 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant42, 0), kwargs = {})
%unsqueeze_48 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_47, 2), kwargs = {})
%unsqueeze_49 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_48, 3), kwargs = {})
%mul_16 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_13, %unsqueeze_49), kwargs = {})
%add_20 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_16, %unsqueeze_46), kwargs = {})
%convert_element_type_25 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_20, torch.float16), kwargs = {})
%convert_element_type_26 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_13, torch.float16), kwargs = {})
%convert_element_type_27 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_6, torch.float16), kwargs = {})
%squeeze_24 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_26, 3), kwargs = {})
%squeeze_25 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_24, 2), kwargs = {})
%squeeze_26 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_27, 3), kwargs = {})
%squeeze_27 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_26, 2), kwargs = {})
%detach_12 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_25,), kwargs = {})
%detach_13 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_27,), kwargs = {})
%permute : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%convert_element_type_25, [0, 2, 3, 1]), kwargs = {})
%view_14 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%permute, [2, 4096, 512]), kwargs = {})
%_param_constant43 : [#users=1] = get_attr[target=_param_constant43]
%t_5 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant43,), kwargs = {})
%expand_1 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%view_14, [2, 4096, 512]), kwargs = {})
%view_15 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_1, [2, 4096, 512]), kwargs = {})
%expand_2 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%t_5, [2, 512, 512]), kwargs = {})
%view_16 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_2, [2, 512, 512]), kwargs = {})
%bmm : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%view_15, %view_16), kwargs = {})
%_unsafe_view : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%bmm, [2, 4096, 512]), kwargs = {})
%_param_constant44 : [#users=1] = get_attr[target=_param_constant44]
%add_21 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%_unsafe_view, %_param_constant44), kwargs = {})
%convert_element_type_28 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_21, torch.float32), kwargs = {})
%var_mean_7 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_28, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_14 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_7, 0), kwargs = {})
%getitem_15 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_7, 1), kwargs = {})
%add_22 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_14, 1e-05), kwargs = {})
%rsqrt_7 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_22,), kwargs = {})
%sub_7 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_21, %getitem_15), kwargs = {})
%mul_17 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_7, %rsqrt_7), kwargs = {})
%_param_constant45 : [#users=1] = get_attr[target=_param_constant45]
%mul_18 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_17, %_param_constant45), kwargs = {})
%_param_constant46 : [#users=1] = get_attr[target=_param_constant46]
%add_23 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_18, %_param_constant46), kwargs = {})
%convert_element_type_29 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_23, torch.float16), kwargs = {})
%_param_constant47 : [#users=1] = get_attr[target=_param_constant47]
%t_6 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant47,), kwargs = {})
%view_17 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_29, [8192, 512]), kwargs = {})
%mm : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_17, %t_6), kwargs = {})
%_unsafe_view_1 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm, [2, 4096, 512]), kwargs = {})
%view_18 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_1, [2, 4096, 8, 64]), kwargs = {})
%permute_1 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_18, [0, 2, 1, 3]), kwargs = {})
%clone : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_1,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_2 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone, [16, 4096, 64]), kwargs = {})
%_param_constant48 : [#users=1] = get_attr[target=_param_constant48]
%t_7 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant48,), kwargs = {})
%view_19 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_1 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_19, %t_7), kwargs = {})
%_unsafe_view_3 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_1, [2, 77, 512]), kwargs = {})
%_param_constant49 : [#users=1] = get_attr[target=_param_constant49]
%t_8 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant49,), kwargs = {})
%view_20 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_2 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_20, %t_8), kwargs = {})
%_unsafe_view_4 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_2, [2, 77, 512]), kwargs = {})
%view_21 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_3, [2, 77, 8, 64]), kwargs = {})
%permute_2 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_21, [0, 2, 1, 3]), kwargs = {})
%clone_1 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_2,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_5 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_1, [16, 77, 64]), kwargs = {})
%view_22 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_4, [2, 77, 8, 64]), kwargs = {})
%permute_3 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_22, [0, 2, 1, 3]), kwargs = {})
%clone_2 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_3,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_6 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_2, [16, 77, 64]), kwargs = {})
%empty : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 4096, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_5, -1, -2), kwargs = {})
%baddbmm : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty, %_unsafe_view_2, %transpose), kwargs = {beta: 0, alpha: 0.125})
%_softmax : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm, -1, False), kwargs = {})
%detach_14 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax,), kwargs = {})
%bmm_1 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax, %_unsafe_view_6), kwargs = {})
%view_23 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_1, [2, 8, 4096, 64]), kwargs = {})
%permute_4 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_23, [0, 2, 1, 3]), kwargs = {})
%clone_3 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_4,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_7 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_3, [2, 4096, 512]), kwargs = {})
%_param_constant50 : [#users=1] = get_attr[target=_param_constant50]
%t_9 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant50,), kwargs = {})
%view_24 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_7, [8192, 512]), kwargs = {})
%_param_constant51 : [#users=1] = get_attr[target=_param_constant51]
%addmm_5 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant51, %view_24, %t_9), kwargs = {})
%view_25 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_5, [2, 4096, 512]), kwargs = {})
%add_24 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_25, %add_21), kwargs = {})
%convert_element_type_30 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_24, torch.float32), kwargs = {})
%var_mean_8 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_30, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_16 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_8, 0), kwargs = {})
%getitem_17 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_8, 1), kwargs = {})
%add_25 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_16, 1e-05), kwargs = {})
%rsqrt_8 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_25,), kwargs = {})
%sub_8 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_24, %getitem_17), kwargs = {})
%mul_19 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_8, %rsqrt_8), kwargs = {})
%_param_constant52 : [#users=1] = get_attr[target=_param_constant52]
%mul_20 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_19, %_param_constant52), kwargs = {})
%_param_constant53 : [#users=1] = get_attr[target=_param_constant53]
%add_26 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_20, %_param_constant53), kwargs = {})
%convert_element_type_31 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_26, torch.float16), kwargs = {})
%_param_constant54 : [#users=1] = get_attr[target=_param_constant54]
%t_10 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant54,), kwargs = {})
%view_26 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_31, [8192, 512]), kwargs = {})
%mm_3 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_26, %t_10), kwargs = {})
%_unsafe_view_8 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_3, [2, 4096, 512]), kwargs = {})
%view_27 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_8, [2, 4096, 8, 64]), kwargs = {})
%permute_5 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_27, [0, 2, 1, 3]), kwargs = {})
%clone_4 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_5,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_9 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_4, [16, 4096, 64]), kwargs = {})
%_param_constant55 : [#users=1] = get_attr[target=_param_constant55]
%t_11 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant55,), kwargs = {})
%view_28 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_4 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_28, %t_11), kwargs = {})
%_unsafe_view_10 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_4, [2, 77, 512]), kwargs = {})
%_param_constant56 : [#users=1] = get_attr[target=_param_constant56]
%t_12 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant56,), kwargs = {})
%view_29 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_5 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_29, %t_12), kwargs = {})
%_unsafe_view_11 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_5, [2, 77, 512]), kwargs = {})
%view_30 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_10, [2, 77, 8, 64]), kwargs = {})
%permute_6 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_30, [0, 2, 1, 3]), kwargs = {})
%clone_5 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_6,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_12 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_5, [16, 77, 64]), kwargs = {})
%view_31 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_11, [2, 77, 8, 64]), kwargs = {})
%permute_7 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_31, [0, 2, 1, 3]), kwargs = {})
%clone_6 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_7,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_13 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_6, [16, 77, 64]), kwargs = {})
%empty_1 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 4096, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_1 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_12, -1, -2), kwargs = {})
%baddbmm_1 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_1, %_unsafe_view_9, %transpose_1), kwargs = {beta: 0, alpha: 0.125})
%_softmax_1 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_1, -1, False), kwargs = {})
%detach_15 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_1,), kwargs = {})
%bmm_2 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_1, %_unsafe_view_13), kwargs = {})
%view_32 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_2, [2, 8, 4096, 64]), kwargs = {})
%permute_8 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_32, [0, 2, 1, 3]), kwargs = {})
%clone_7 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_8,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_14 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_7, [2, 4096, 512]), kwargs = {})
%_param_constant57 : [#users=1] = get_attr[target=_param_constant57]
%t_13 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant57,), kwargs = {})
%view_33 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_14, [8192, 512]), kwargs = {})
%_param_constant58 : [#users=1] = get_attr[target=_param_constant58]
%addmm_6 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant58, %view_33, %t_13), kwargs = {})
%view_34 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_6, [2, 4096, 512]), kwargs = {})
%add_27 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_34, %add_24), kwargs = {})
%convert_element_type_32 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_27, torch.float32), kwargs = {})
%var_mean_9 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_32, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_18 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_9, 0), kwargs = {})
%getitem_19 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_9, 1), kwargs = {})
%add_28 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_18, 1e-05), kwargs = {})
%rsqrt_9 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_28,), kwargs = {})
%sub_9 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_27, %getitem_19), kwargs = {})
%mul_21 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_9, %rsqrt_9), kwargs = {})
%_param_constant59 : [#users=1] = get_attr[target=_param_constant59]
%mul_22 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_21, %_param_constant59), kwargs = {})
%_param_constant60 : [#users=1] = get_attr[target=_param_constant60]
%add_29 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_22, %_param_constant60), kwargs = {})
%convert_element_type_33 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_29, torch.float16), kwargs = {})
%_param_constant61 : [#users=1] = get_attr[target=_param_constant61]
%t_14 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant61,), kwargs = {})
%view_35 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_33, [8192, 512]), kwargs = {})
%_param_constant62 : [#users=1] = get_attr[target=_param_constant62]
%addmm_7 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant62, %view_35, %t_14), kwargs = {})
%view_36 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%addmm_7, [2, 4096, 4096]), kwargs = {})
%slice_13 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_36, -1, 0, 2048), kwargs = {})
%slice_14 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_36, -1, 2048, 4096), kwargs = {})
%gelu : [#users=1] = call_function[target=torch.ops.aten.gelu](args = (%slice_14,), kwargs = {})
%mul_23 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%slice_13, %gelu), kwargs = {})
%_param_constant63 : [#users=1] = get_attr[target=_param_constant63]
%t_15 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant63,), kwargs = {})
%view_37 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_23, [8192, 2048]), kwargs = {})
%_param_constant64 : [#users=1] = get_attr[target=_param_constant64]
%addmm_8 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant64, %view_37, %t_15), kwargs = {})
%view_38 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_8, [2, 4096, 512]), kwargs = {})
%add_30 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%view_38, %add_27), kwargs = {})
%_param_constant65 : [#users=1] = get_attr[target=_param_constant65]
%t_16 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant65,), kwargs = {})
%view_39 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%add_30, [8192, 512]), kwargs = {})
%_param_constant66 : [#users=1] = get_attr[target=_param_constant66]
%addmm_9 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant66, %view_39, %t_16), kwargs = {})
%view_40 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_9, [2, 4096, 512]), kwargs = {})
%view_41 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%view_40, [2, 64, 64, 512]), kwargs = {})
%permute_9 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_41, [0, 3, 1, 2]), kwargs = {})
%clone_8 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_9,), kwargs = {memory_format: torch.contiguous_format})
%add_31 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%clone_8, %div_3), kwargs = {})
%view_42 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_31, [2, 32, 16, 4096]), kwargs = {})
%convert_element_type_34 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_42, torch.float32), kwargs = {})
%var_mean_10 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_34, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_20 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_10, 0), kwargs = {})
%getitem_21 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_10, 1), kwargs = {})
%add_32 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_20, 1e-05), kwargs = {})
%rsqrt_10 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_32,), kwargs = {})
%sub_10 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_42, %getitem_21), kwargs = {})
%mul_24 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_10, %rsqrt_10), kwargs = {})
%view_43 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_24, [2, 512, 64, 64]), kwargs = {})
%_param_constant67 : [#users=1] = get_attr[target=_param_constant67]
%unsqueeze_50 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant67, 0), kwargs = {})
%unsqueeze_51 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_50, 2), kwargs = {})
%unsqueeze_52 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_51, 3), kwargs = {})
%_param_constant68 : [#users=1] = get_attr[target=_param_constant68]
%unsqueeze_53 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant68, 0), kwargs = {})
%unsqueeze_54 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_53, 2), kwargs = {})
%unsqueeze_55 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_54, 3), kwargs = {})
%mul_25 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_43, %unsqueeze_55), kwargs = {})
%add_33 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_25, %unsqueeze_52), kwargs = {})
%convert_element_type_35 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_33, torch.float16), kwargs = {})
%convert_element_type_36 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_21, torch.float16), kwargs = {})
%convert_element_type_37 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_10, torch.float16), kwargs = {})
%squeeze_28 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_36, 3), kwargs = {})
%squeeze_29 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_28, 2), kwargs = {})
%squeeze_30 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_37, 3), kwargs = {})
%squeeze_31 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_30, 2), kwargs = {})
%detach_16 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_29,), kwargs = {})
%detach_17 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_31,), kwargs = {})
%silu_10 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_35,), kwargs = {})
%_param_constant69 : [#users=1] = get_attr[target=_param_constant69]
%_param_constant70 : [#users=1] = get_attr[target=_param_constant70]
%convolution_9 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_10, %_param_constant69, %_param_constant70, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_11 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant71 : [#users=1] = get_attr[target=_param_constant71]
%t_17 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant71,), kwargs = {})
%_param_constant72 : [#users=1] = get_attr[target=_param_constant72]
%addmm_10 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant72, %silu_11, %t_17), kwargs = {})
%slice_15 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_10, 0, 0, 9223372036854775807), kwargs = {})
%slice_16 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_15, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_56 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_16, 2), kwargs = {})
%unsqueeze_57 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_56, 3), kwargs = {})
%add_34 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_9, %unsqueeze_57), kwargs = {})
%view_44 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_34, [2, 32, 16, 4096]), kwargs = {})
%convert_element_type_38 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_44, torch.float32), kwargs = {})
%var_mean_11 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_38, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_22 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_11, 0), kwargs = {})
%getitem_23 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_11, 1), kwargs = {})
%add_35 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_22, 1e-05), kwargs = {})
%rsqrt_11 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_35,), kwargs = {})
%sub_11 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_44, %getitem_23), kwargs = {})
%mul_26 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_11, %rsqrt_11), kwargs = {})
%view_45 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_26, [2, 512, 64, 64]), kwargs = {})
%_param_constant73 : [#users=1] = get_attr[target=_param_constant73]
%unsqueeze_58 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant73, 0), kwargs = {})
%unsqueeze_59 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_58, 2), kwargs = {})
%unsqueeze_60 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_59, 3), kwargs = {})
%_param_constant74 : [#users=1] = get_attr[target=_param_constant74]
%unsqueeze_61 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant74, 0), kwargs = {})
%unsqueeze_62 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_61, 2), kwargs = {})
%unsqueeze_63 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_62, 3), kwargs = {})
%mul_27 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_45, %unsqueeze_63), kwargs = {})
%add_36 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_27, %unsqueeze_60), kwargs = {})
%convert_element_type_39 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_36, torch.float16), kwargs = {})
%convert_element_type_40 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_23, torch.float16), kwargs = {})
%convert_element_type_41 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_11, torch.float16), kwargs = {})
%squeeze_32 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_40, 3), kwargs = {})
%squeeze_33 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_32, 2), kwargs = {})
%squeeze_34 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_41, 3), kwargs = {})
%squeeze_35 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_34, 2), kwargs = {})
%detach_18 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_33,), kwargs = {})
%detach_19 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_35,), kwargs = {})
%silu_12 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_39,), kwargs = {})
%_param_constant75 : [#users=1] = get_attr[target=_param_constant75]
%_param_constant76 : [#users=1] = get_attr[target=_param_constant76]
%convolution_10 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_12, %_param_constant75, %_param_constant76, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%add_37 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%add_31, %convolution_10), kwargs = {})
%div_4 : [#users=2] = call_function[target=torch.ops.aten.div](args = (%add_37, 1.0), kwargs = {})
%view_46 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%div_4, [2, 32, 16, 4096]), kwargs = {})
%convert_element_type_42 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_46, torch.float32), kwargs = {})
%var_mean_12 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_42, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_24 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_12, 0), kwargs = {})
%getitem_25 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_12, 1), kwargs = {})
%add_38 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_24, 1e-06), kwargs = {})
%rsqrt_12 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_38,), kwargs = {})
%sub_12 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_46, %getitem_25), kwargs = {})
%mul_28 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_12, %rsqrt_12), kwargs = {})
%view_47 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_28, [2, 512, 64, 64]), kwargs = {})
%_param_constant77 : [#users=1] = get_attr[target=_param_constant77]
%unsqueeze_64 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant77, 0), kwargs = {})
%unsqueeze_65 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_64, 2), kwargs = {})
%unsqueeze_66 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_65, 3), kwargs = {})
%_param_constant78 : [#users=1] = get_attr[target=_param_constant78]
%unsqueeze_67 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant78, 0), kwargs = {})
%unsqueeze_68 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_67, 2), kwargs = {})
%unsqueeze_69 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_68, 3), kwargs = {})
%mul_29 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_47, %unsqueeze_69), kwargs = {})
%add_39 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_29, %unsqueeze_66), kwargs = {})
%convert_element_type_43 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_39, torch.float16), kwargs = {})
%convert_element_type_44 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_25, torch.float16), kwargs = {})
%convert_element_type_45 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_12, torch.float16), kwargs = {})
%squeeze_36 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_44, 3), kwargs = {})
%squeeze_37 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_36, 2), kwargs = {})
%squeeze_38 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_45, 3), kwargs = {})
%squeeze_39 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_38, 2), kwargs = {})
%detach_20 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_37,), kwargs = {})
%detach_21 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_39,), kwargs = {})
%permute_10 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%convert_element_type_43, [0, 2, 3, 1]), kwargs = {})
%view_48 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%permute_10, [2, 4096, 512]), kwargs = {})
%_param_constant79 : [#users=1] = get_attr[target=_param_constant79]
%t_18 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant79,), kwargs = {})
%expand_3 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%view_48, [2, 4096, 512]), kwargs = {})
%view_49 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_3, [2, 4096, 512]), kwargs = {})
%expand_4 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%t_18, [2, 512, 512]), kwargs = {})
%view_50 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_4, [2, 512, 512]), kwargs = {})
%bmm_3 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%view_49, %view_50), kwargs = {})
%_unsafe_view_15 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%bmm_3, [2, 4096, 512]), kwargs = {})
%_param_constant80 : [#users=1] = get_attr[target=_param_constant80]
%add_40 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%_unsafe_view_15, %_param_constant80), kwargs = {})
%convert_element_type_46 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_40, torch.float32), kwargs = {})
%var_mean_13 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_46, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_26 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_13, 0), kwargs = {})
%getitem_27 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_13, 1), kwargs = {})
%add_41 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_26, 1e-05), kwargs = {})
%rsqrt_13 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_41,), kwargs = {})
%sub_13 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_40, %getitem_27), kwargs = {})
%mul_30 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_13, %rsqrt_13), kwargs = {})
%_param_constant81 : [#users=1] = get_attr[target=_param_constant81]
%mul_31 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_30, %_param_constant81), kwargs = {})
%_param_constant82 : [#users=1] = get_attr[target=_param_constant82]
%add_42 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_31, %_param_constant82), kwargs = {})
%convert_element_type_47 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_42, torch.float16), kwargs = {})
%_param_constant83 : [#users=1] = get_attr[target=_param_constant83]
%t_19 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant83,), kwargs = {})
%view_51 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_47, [8192, 512]), kwargs = {})
%mm_6 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_51, %t_19), kwargs = {})
%_unsafe_view_16 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_6, [2, 4096, 512]), kwargs = {})
%view_52 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_16, [2, 4096, 8, 64]), kwargs = {})
%permute_11 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_52, [0, 2, 1, 3]), kwargs = {})
%clone_9 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_11,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_17 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_9, [16, 4096, 64]), kwargs = {})
%_param_constant84 : [#users=1] = get_attr[target=_param_constant84]
%t_20 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant84,), kwargs = {})
%view_53 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_7 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_53, %t_20), kwargs = {})
%_unsafe_view_18 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_7, [2, 77, 512]), kwargs = {})
%_param_constant85 : [#users=1] = get_attr[target=_param_constant85]
%t_21 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant85,), kwargs = {})
%view_54 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_8 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_54, %t_21), kwargs = {})
%_unsafe_view_19 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_8, [2, 77, 512]), kwargs = {})
%view_55 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_18, [2, 77, 8, 64]), kwargs = {})
%permute_12 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_55, [0, 2, 1, 3]), kwargs = {})
%clone_10 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_12,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_20 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_10, [16, 77, 64]), kwargs = {})
%view_56 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_19, [2, 77, 8, 64]), kwargs = {})
%permute_13 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_56, [0, 2, 1, 3]), kwargs = {})
%clone_11 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_13,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_21 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_11, [16, 77, 64]), kwargs = {})
%empty_2 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 4096, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_2 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_20, -1, -2), kwargs = {})
%baddbmm_2 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_2, %_unsafe_view_17, %transpose_2), kwargs = {beta: 0, alpha: 0.125})
%_softmax_2 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_2, -1, False), kwargs = {})
%detach_22 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_2,), kwargs = {})
%bmm_4 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_2, %_unsafe_view_21), kwargs = {})
%view_57 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_4, [2, 8, 4096, 64]), kwargs = {})
%permute_14 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_57, [0, 2, 1, 3]), kwargs = {})
%clone_12 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_14,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_22 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_12, [2, 4096, 512]), kwargs = {})
%_param_constant86 : [#users=1] = get_attr[target=_param_constant86]
%t_22 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant86,), kwargs = {})
%view_58 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_22, [8192, 512]), kwargs = {})
%_param_constant87 : [#users=1] = get_attr[target=_param_constant87]
%addmm_11 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant87, %view_58, %t_22), kwargs = {})
%view_59 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_11, [2, 4096, 512]), kwargs = {})
%add_43 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_59, %add_40), kwargs = {})
%convert_element_type_48 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_43, torch.float32), kwargs = {})
%var_mean_14 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_48, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_28 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_14, 0), kwargs = {})
%getitem_29 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_14, 1), kwargs = {})
%add_44 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_28, 1e-05), kwargs = {})
%rsqrt_14 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_44,), kwargs = {})
%sub_14 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_43, %getitem_29), kwargs = {})
%mul_32 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_14, %rsqrt_14), kwargs = {})
%_param_constant88 : [#users=1] = get_attr[target=_param_constant88]
%mul_33 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_32, %_param_constant88), kwargs = {})
%_param_constant89 : [#users=1] = get_attr[target=_param_constant89]
%add_45 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_33, %_param_constant89), kwargs = {})
%convert_element_type_49 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_45, torch.float16), kwargs = {})
%_param_constant90 : [#users=1] = get_attr[target=_param_constant90]
%t_23 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant90,), kwargs = {})
%view_60 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_49, [8192, 512]), kwargs = {})
%mm_9 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_60, %t_23), kwargs = {})
%_unsafe_view_23 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_9, [2, 4096, 512]), kwargs = {})
%view_61 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_23, [2, 4096, 8, 64]), kwargs = {})
%permute_15 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_61, [0, 2, 1, 3]), kwargs = {})
%clone_13 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_15,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_24 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_13, [16, 4096, 64]), kwargs = {})
%_param_constant91 : [#users=1] = get_attr[target=_param_constant91]
%t_24 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant91,), kwargs = {})
%view_62 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_10 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_62, %t_24), kwargs = {})
%_unsafe_view_25 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_10, [2, 77, 512]), kwargs = {})
%_param_constant92 : [#users=1] = get_attr[target=_param_constant92]
%t_25 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant92,), kwargs = {})
%view_63 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_11 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_63, %t_25), kwargs = {})
%_unsafe_view_26 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_11, [2, 77, 512]), kwargs = {})
%view_64 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_25, [2, 77, 8, 64]), kwargs = {})
%permute_16 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_64, [0, 2, 1, 3]), kwargs = {})
%clone_14 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_16,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_27 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_14, [16, 77, 64]), kwargs = {})
%view_65 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_26, [2, 77, 8, 64]), kwargs = {})
%permute_17 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_65, [0, 2, 1, 3]), kwargs = {})
%clone_15 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_17,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_28 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_15, [16, 77, 64]), kwargs = {})
%empty_3 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 4096, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_3 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_27, -1, -2), kwargs = {})
%baddbmm_3 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_3, %_unsafe_view_24, %transpose_3), kwargs = {beta: 0, alpha: 0.125})
%_softmax_3 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_3, -1, False), kwargs = {})
%detach_23 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_3,), kwargs = {})
%bmm_5 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_3, %_unsafe_view_28), kwargs = {})
%view_66 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_5, [2, 8, 4096, 64]), kwargs = {})
%permute_18 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_66, [0, 2, 1, 3]), kwargs = {})
%clone_16 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_18,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_29 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_16, [2, 4096, 512]), kwargs = {})
%_param_constant93 : [#users=1] = get_attr[target=_param_constant93]
%t_26 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant93,), kwargs = {})
%view_67 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_29, [8192, 512]), kwargs = {})
%_param_constant94 : [#users=1] = get_attr[target=_param_constant94]
%addmm_12 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant94, %view_67, %t_26), kwargs = {})
%view_68 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_12, [2, 4096, 512]), kwargs = {})
%add_46 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_68, %add_43), kwargs = {})
%convert_element_type_50 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_46, torch.float32), kwargs = {})
%var_mean_15 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_50, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_30 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_15, 0), kwargs = {})
%getitem_31 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_15, 1), kwargs = {})
%add_47 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_30, 1e-05), kwargs = {})
%rsqrt_15 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_47,), kwargs = {})
%sub_15 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_46, %getitem_31), kwargs = {})
%mul_34 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_15, %rsqrt_15), kwargs = {})
%_param_constant95 : [#users=1] = get_attr[target=_param_constant95]
%mul_35 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_34, %_param_constant95), kwargs = {})
%_param_constant96 : [#users=1] = get_attr[target=_param_constant96]
%add_48 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_35, %_param_constant96), kwargs = {})
%convert_element_type_51 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_48, torch.float16), kwargs = {})
%_param_constant97 : [#users=1] = get_attr[target=_param_constant97]
%t_27 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant97,), kwargs = {})
%view_69 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_51, [8192, 512]), kwargs = {})
%_param_constant98 : [#users=1] = get_attr[target=_param_constant98]
%addmm_13 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant98, %view_69, %t_27), kwargs = {})
%view_70 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%addmm_13, [2, 4096, 4096]), kwargs = {})
%slice_17 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_70, -1, 0, 2048), kwargs = {})
%slice_18 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_70, -1, 2048, 4096), kwargs = {})
%gelu_1 : [#users=1] = call_function[target=torch.ops.aten.gelu](args = (%slice_18,), kwargs = {})
%mul_36 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%slice_17, %gelu_1), kwargs = {})
%_param_constant99 : [#users=1] = get_attr[target=_param_constant99]
%t_28 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant99,), kwargs = {})
%view_71 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_36, [8192, 2048]), kwargs = {})
%_param_constant100 : [#users=1] = get_attr[target=_param_constant100]
%addmm_14 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant100, %view_71, %t_28), kwargs = {})
%view_72 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_14, [2, 4096, 512]), kwargs = {})
%add_49 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%view_72, %add_46), kwargs = {})
%_param_constant101 : [#users=1] = get_attr[target=_param_constant101]
%t_29 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant101,), kwargs = {})
%view_73 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%add_49, [8192, 512]), kwargs = {})
%_param_constant102 : [#users=1] = get_attr[target=_param_constant102]
%addmm_15 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant102, %view_73, %t_29), kwargs = {})
%view_74 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_15, [2, 4096, 512]), kwargs = {})
%view_75 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%view_74, [2, 64, 64, 512]), kwargs = {})
%permute_19 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_75, [0, 3, 1, 2]), kwargs = {})
%clone_17 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_19,), kwargs = {memory_format: torch.contiguous_format})
%add_50 : [#users=2] = call_function[target=torch.ops.aten.add](args = (%clone_17, %div_4), kwargs = {})
%_param_constant103 : [#users=1] = get_attr[target=_param_constant103]
%_param_constant104 : [#users=1] = get_attr[target=_param_constant104]
%convolution_11 : [#users=3] = call_function[target=torch.ops.aten.convolution](args = (%add_50, %_param_constant103, %_param_constant104, [2, 2], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%view_76 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%convolution_11, [2, 32, 16, 1024]), kwargs = {})
%convert_element_type_52 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_76, torch.float32), kwargs = {})
%var_mean_16 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_52, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_32 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_16, 0), kwargs = {})
%getitem_33 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_16, 1), kwargs = {})
%add_51 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_32, 1e-05), kwargs = {})
%rsqrt_16 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_51,), kwargs = {})
%sub_16 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_76, %getitem_33), kwargs = {})
%mul_37 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_16, %rsqrt_16), kwargs = {})
%view_77 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_37, [2, 512, 32, 32]), kwargs = {})
%_param_constant105 : [#users=1] = get_attr[target=_param_constant105]
%unsqueeze_70 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant105, 0), kwargs = {})
%unsqueeze_71 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_70, 2), kwargs = {})
%unsqueeze_72 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_71, 3), kwargs = {})
%_param_constant106 : [#users=1] = get_attr[target=_param_constant106]
%unsqueeze_73 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant106, 0), kwargs = {})
%unsqueeze_74 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_73, 2), kwargs = {})
%unsqueeze_75 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_74, 3), kwargs = {})
%mul_38 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_77, %unsqueeze_75), kwargs = {})
%add_52 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_38, %unsqueeze_72), kwargs = {})
%convert_element_type_53 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_52, torch.float16), kwargs = {})
%convert_element_type_54 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_33, torch.float16), kwargs = {})
%convert_element_type_55 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_16, torch.float16), kwargs = {})
%squeeze_40 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_54, 3), kwargs = {})
%squeeze_41 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_40, 2), kwargs = {})
%squeeze_42 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_55, 3), kwargs = {})
%squeeze_43 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_42, 2), kwargs = {})
%detach_24 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_41,), kwargs = {})
%detach_25 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_43,), kwargs = {})
%silu_13 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_53,), kwargs = {})
%_param_constant107 : [#users=1] = get_attr[target=_param_constant107]
%_param_constant108 : [#users=1] = get_attr[target=_param_constant108]
%convolution_12 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_13, %_param_constant107, %_param_constant108, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_14 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant109 : [#users=1] = get_attr[target=_param_constant109]
%t_30 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant109,), kwargs = {})
%_param_constant110 : [#users=1] = get_attr[target=_param_constant110]
%addmm_16 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant110, %silu_14, %t_30), kwargs = {})
%slice_19 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_16, 0, 0, 9223372036854775807), kwargs = {})
%slice_20 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_19, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_76 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_20, 2), kwargs = {})
%unsqueeze_77 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_76, 3), kwargs = {})
%add_53 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_12, %unsqueeze_77), kwargs = {})
%view_78 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_53, [2, 32, 16, 1024]), kwargs = {})
%convert_element_type_56 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_78, torch.float32), kwargs = {})
%var_mean_17 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_56, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_34 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_17, 0), kwargs = {})
%getitem_35 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_17, 1), kwargs = {})
%add_54 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_34, 1e-05), kwargs = {})
%rsqrt_17 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_54,), kwargs = {})
%sub_17 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_78, %getitem_35), kwargs = {})
%mul_39 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_17, %rsqrt_17), kwargs = {})
%view_79 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_39, [2, 512, 32, 32]), kwargs = {})
%_param_constant111 : [#users=1] = get_attr[target=_param_constant111]
%unsqueeze_78 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant111, 0), kwargs = {})
%unsqueeze_79 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_78, 2), kwargs = {})
%unsqueeze_80 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_79, 3), kwargs = {})
%_param_constant112 : [#users=1] = get_attr[target=_param_constant112]
%unsqueeze_81 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant112, 0), kwargs = {})
%unsqueeze_82 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_81, 2), kwargs = {})
%unsqueeze_83 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_82, 3), kwargs = {})
%mul_40 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_79, %unsqueeze_83), kwargs = {})
%add_55 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_40, %unsqueeze_80), kwargs = {})
%convert_element_type_57 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_55, torch.float16), kwargs = {})
%convert_element_type_58 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_35, torch.float16), kwargs = {})
%convert_element_type_59 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_17, torch.float16), kwargs = {})
%squeeze_44 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_58, 3), kwargs = {})
%squeeze_45 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_44, 2), kwargs = {})
%squeeze_46 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_59, 3), kwargs = {})
%squeeze_47 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_46, 2), kwargs = {})
%detach_26 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_45,), kwargs = {})
%detach_27 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_47,), kwargs = {})
%silu_15 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_57,), kwargs = {})
%_param_constant113 : [#users=1] = get_attr[target=_param_constant113]
%_param_constant114 : [#users=1] = get_attr[target=_param_constant114]
%convolution_13 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_15, %_param_constant113, %_param_constant114, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%add_56 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_11, %convolution_13), kwargs = {})
%div_5 : [#users=2] = call_function[target=torch.ops.aten.div](args = (%add_56, 1.0), kwargs = {})
%view_80 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%div_5, [2, 32, 16, 1024]), kwargs = {})
%convert_element_type_60 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_80, torch.float32), kwargs = {})
%var_mean_18 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_60, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_36 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_18, 0), kwargs = {})
%getitem_37 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_18, 1), kwargs = {})
%add_57 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_36, 1e-06), kwargs = {})
%rsqrt_18 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_57,), kwargs = {})
%sub_18 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_80, %getitem_37), kwargs = {})
%mul_41 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_18, %rsqrt_18), kwargs = {})
%view_81 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_41, [2, 512, 32, 32]), kwargs = {})
%_param_constant115 : [#users=1] = get_attr[target=_param_constant115]
%unsqueeze_84 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant115, 0), kwargs = {})
%unsqueeze_85 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_84, 2), kwargs = {})
%unsqueeze_86 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_85, 3), kwargs = {})
%_param_constant116 : [#users=1] = get_attr[target=_param_constant116]
%unsqueeze_87 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant116, 0), kwargs = {})
%unsqueeze_88 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_87, 2), kwargs = {})
%unsqueeze_89 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_88, 3), kwargs = {})
%mul_42 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_81, %unsqueeze_89), kwargs = {})
%add_58 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_42, %unsqueeze_86), kwargs = {})
%convert_element_type_61 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_58, torch.float16), kwargs = {})
%convert_element_type_62 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_37, torch.float16), kwargs = {})
%convert_element_type_63 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_18, torch.float16), kwargs = {})
%squeeze_48 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_62, 3), kwargs = {})
%squeeze_49 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_48, 2), kwargs = {})
%squeeze_50 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_63, 3), kwargs = {})
%squeeze_51 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_50, 2), kwargs = {})
%detach_28 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_49,), kwargs = {})
%detach_29 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_51,), kwargs = {})
%permute_20 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%convert_element_type_61, [0, 2, 3, 1]), kwargs = {})
%view_82 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%permute_20, [2, 1024, 512]), kwargs = {})
%_param_constant117 : [#users=1] = get_attr[target=_param_constant117]
%t_31 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant117,), kwargs = {})
%expand_5 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%view_82, [2, 1024, 512]), kwargs = {})
%view_83 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_5, [2, 1024, 512]), kwargs = {})
%expand_6 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%t_31, [2, 512, 512]), kwargs = {})
%view_84 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_6, [2, 512, 512]), kwargs = {})
%bmm_6 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%view_83, %view_84), kwargs = {})
%_unsafe_view_30 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%bmm_6, [2, 1024, 512]), kwargs = {})
%_param_constant118 : [#users=1] = get_attr[target=_param_constant118]
%add_59 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%_unsafe_view_30, %_param_constant118), kwargs = {})
%convert_element_type_64 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_59, torch.float32), kwargs = {})
%var_mean_19 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_64, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_38 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_19, 0), kwargs = {})
%getitem_39 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_19, 1), kwargs = {})
%add_60 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_38, 1e-05), kwargs = {})
%rsqrt_19 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_60,), kwargs = {})
%sub_19 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_59, %getitem_39), kwargs = {})
%mul_43 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_19, %rsqrt_19), kwargs = {})
%_param_constant119 : [#users=1] = get_attr[target=_param_constant119]
%mul_44 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_43, %_param_constant119), kwargs = {})
%_param_constant120 : [#users=1] = get_attr[target=_param_constant120]
%add_61 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_44, %_param_constant120), kwargs = {})
%convert_element_type_65 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_61, torch.float16), kwargs = {})
%_param_constant121 : [#users=1] = get_attr[target=_param_constant121]
%t_32 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant121,), kwargs = {})
%view_85 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_65, [2048, 512]), kwargs = {})
%mm_12 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_85, %t_32), kwargs = {})
%_unsafe_view_31 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_12, [2, 1024, 512]), kwargs = {})
%view_86 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_31, [2, 1024, 8, 64]), kwargs = {})
%permute_21 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_86, [0, 2, 1, 3]), kwargs = {})
%clone_18 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_21,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_32 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_18, [16, 1024, 64]), kwargs = {})
%_param_constant122 : [#users=1] = get_attr[target=_param_constant122]
%t_33 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant122,), kwargs = {})
%view_87 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_13 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_87, %t_33), kwargs = {})
%_unsafe_view_33 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_13, [2, 77, 512]), kwargs = {})
%_param_constant123 : [#users=1] = get_attr[target=_param_constant123]
%t_34 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant123,), kwargs = {})
%view_88 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_14 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_88, %t_34), kwargs = {})
%_unsafe_view_34 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_14, [2, 77, 512]), kwargs = {})
%view_89 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_33, [2, 77, 8, 64]), kwargs = {})
%permute_22 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_89, [0, 2, 1, 3]), kwargs = {})
%clone_19 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_22,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_35 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_19, [16, 77, 64]), kwargs = {})
%view_90 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_34, [2, 77, 8, 64]), kwargs = {})
%permute_23 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_90, [0, 2, 1, 3]), kwargs = {})
%clone_20 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_23,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_36 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_20, [16, 77, 64]), kwargs = {})
%empty_4 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 1024, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_4 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_35, -1, -2), kwargs = {})
%baddbmm_4 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_4, %_unsafe_view_32, %transpose_4), kwargs = {beta: 0, alpha: 0.125})
%_softmax_4 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_4, -1, False), kwargs = {})
%detach_30 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_4,), kwargs = {})
%bmm_7 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_4, %_unsafe_view_36), kwargs = {})
%view_91 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_7, [2, 8, 1024, 64]), kwargs = {})
%permute_24 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_91, [0, 2, 1, 3]), kwargs = {})
%clone_21 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_24,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_37 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_21, [2, 1024, 512]), kwargs = {})
%_param_constant124 : [#users=1] = get_attr[target=_param_constant124]
%t_35 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant124,), kwargs = {})
%view_92 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_37, [2048, 512]), kwargs = {})
%_param_constant125 : [#users=1] = get_attr[target=_param_constant125]
%addmm_17 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant125, %view_92, %t_35), kwargs = {})
%view_93 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_17, [2, 1024, 512]), kwargs = {})
%add_62 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_93, %add_59), kwargs = {})
%convert_element_type_66 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_62, torch.float32), kwargs = {})
%var_mean_20 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_66, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_40 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_20, 0), kwargs = {})
%getitem_41 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_20, 1), kwargs = {})
%add_63 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_40, 1e-05), kwargs = {})
%rsqrt_20 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_63,), kwargs = {})
%sub_20 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_62, %getitem_41), kwargs = {})
%mul_45 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_20, %rsqrt_20), kwargs = {})
%_param_constant126 : [#users=1] = get_attr[target=_param_constant126]
%mul_46 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_45, %_param_constant126), kwargs = {})
%_param_constant127 : [#users=1] = get_attr[target=_param_constant127]
%add_64 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_46, %_param_constant127), kwargs = {})
%convert_element_type_67 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_64, torch.float16), kwargs = {})
%_param_constant128 : [#users=1] = get_attr[target=_param_constant128]
%t_36 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant128,), kwargs = {})
%view_94 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_67, [2048, 512]), kwargs = {})
%mm_15 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_94, %t_36), kwargs = {})
%_unsafe_view_38 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_15, [2, 1024, 512]), kwargs = {})
%view_95 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_38, [2, 1024, 8, 64]), kwargs = {})
%permute_25 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_95, [0, 2, 1, 3]), kwargs = {})
%clone_22 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_25,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_39 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_22, [16, 1024, 64]), kwargs = {})
%_param_constant129 : [#users=1] = get_attr[target=_param_constant129]
%t_37 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant129,), kwargs = {})
%view_96 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_16 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_96, %t_37), kwargs = {})
%_unsafe_view_40 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_16, [2, 77, 512]), kwargs = {})
%_param_constant130 : [#users=1] = get_attr[target=_param_constant130]
%t_38 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant130,), kwargs = {})
%view_97 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_17 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_97, %t_38), kwargs = {})
%_unsafe_view_41 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_17, [2, 77, 512]), kwargs = {})
%view_98 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_40, [2, 77, 8, 64]), kwargs = {})
%permute_26 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_98, [0, 2, 1, 3]), kwargs = {})
%clone_23 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_26,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_42 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_23, [16, 77, 64]), kwargs = {})
%view_99 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_41, [2, 77, 8, 64]), kwargs = {})
%permute_27 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_99, [0, 2, 1, 3]), kwargs = {})
%clone_24 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_27,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_43 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_24, [16, 77, 64]), kwargs = {})
%empty_5 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 1024, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_5 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_42, -1, -2), kwargs = {})
%baddbmm_5 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_5, %_unsafe_view_39, %transpose_5), kwargs = {beta: 0, alpha: 0.125})
%_softmax_5 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_5, -1, False), kwargs = {})
%detach_31 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_5,), kwargs = {})
%bmm_8 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_5, %_unsafe_view_43), kwargs = {})
%view_100 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_8, [2, 8, 1024, 64]), kwargs = {})
%permute_28 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_100, [0, 2, 1, 3]), kwargs = {})
%clone_25 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_28,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_44 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_25, [2, 1024, 512]), kwargs = {})
%_param_constant131 : [#users=1] = get_attr[target=_param_constant131]
%t_39 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant131,), kwargs = {})
%view_101 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_44, [2048, 512]), kwargs = {})
%_param_constant132 : [#users=1] = get_attr[target=_param_constant132]
%addmm_18 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant132, %view_101, %t_39), kwargs = {})
%view_102 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_18, [2, 1024, 512]), kwargs = {})
%add_65 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_102, %add_62), kwargs = {})
%convert_element_type_68 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_65, torch.float32), kwargs = {})
%var_mean_21 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_68, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_42 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_21, 0), kwargs = {})
%getitem_43 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_21, 1), kwargs = {})
%add_66 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_42, 1e-05), kwargs = {})
%rsqrt_21 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_66,), kwargs = {})
%sub_21 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_65, %getitem_43), kwargs = {})
%mul_47 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_21, %rsqrt_21), kwargs = {})
%_param_constant133 : [#users=1] = get_attr[target=_param_constant133]
%mul_48 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_47, %_param_constant133), kwargs = {})
%_param_constant134 : [#users=1] = get_attr[target=_param_constant134]
%add_67 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_48, %_param_constant134), kwargs = {})
%convert_element_type_69 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_67, torch.float16), kwargs = {})
%_param_constant135 : [#users=1] = get_attr[target=_param_constant135]
%t_40 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant135,), kwargs = {})
%view_103 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_69, [2048, 512]), kwargs = {})
%_param_constant136 : [#users=1] = get_attr[target=_param_constant136]
%addmm_19 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant136, %view_103, %t_40), kwargs = {})
%view_104 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%addmm_19, [2, 1024, 4096]), kwargs = {})
%slice_21 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_104, -1, 0, 2048), kwargs = {})
%slice_22 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_104, -1, 2048, 4096), kwargs = {})
%gelu_2 : [#users=1] = call_function[target=torch.ops.aten.gelu](args = (%slice_22,), kwargs = {})
%mul_49 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%slice_21, %gelu_2), kwargs = {})
%_param_constant137 : [#users=1] = get_attr[target=_param_constant137]
%t_41 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant137,), kwargs = {})
%view_105 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_49, [2048, 2048]), kwargs = {})
%_param_constant138 : [#users=1] = get_attr[target=_param_constant138]
%addmm_20 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant138, %view_105, %t_41), kwargs = {})
%view_106 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_20, [2, 1024, 512]), kwargs = {})
%add_68 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%view_106, %add_65), kwargs = {})
%_param_constant139 : [#users=1] = get_attr[target=_param_constant139]
%t_42 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant139,), kwargs = {})
%view_107 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%add_68, [2048, 512]), kwargs = {})
%_param_constant140 : [#users=1] = get_attr[target=_param_constant140]
%addmm_21 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant140, %view_107, %t_42), kwargs = {})
%view_108 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_21, [2, 1024, 512]), kwargs = {})
%view_109 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%view_108, [2, 32, 32, 512]), kwargs = {})
%permute_29 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_109, [0, 3, 1, 2]), kwargs = {})
%clone_26 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_29,), kwargs = {memory_format: torch.contiguous_format})
%add_69 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%clone_26, %div_5), kwargs = {})
%view_110 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_69, [2, 32, 16, 1024]), kwargs = {})
%convert_element_type_70 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_110, torch.float32), kwargs = {})
%var_mean_22 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_70, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_44 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_22, 0), kwargs = {})
%getitem_45 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_22, 1), kwargs = {})
%add_70 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_44, 1e-05), kwargs = {})
%rsqrt_22 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_70,), kwargs = {})
%sub_22 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_110, %getitem_45), kwargs = {})
%mul_50 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_22, %rsqrt_22), kwargs = {})
%view_111 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_50, [2, 512, 32, 32]), kwargs = {})
%_param_constant141 : [#users=1] = get_attr[target=_param_constant141]
%unsqueeze_90 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant141, 0), kwargs = {})
%unsqueeze_91 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_90, 2), kwargs = {})
%unsqueeze_92 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_91, 3), kwargs = {})
%_param_constant142 : [#users=1] = get_attr[target=_param_constant142]
%unsqueeze_93 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant142, 0), kwargs = {})
%unsqueeze_94 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_93, 2), kwargs = {})
%unsqueeze_95 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_94, 3), kwargs = {})
%mul_51 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_111, %unsqueeze_95), kwargs = {})
%add_71 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_51, %unsqueeze_92), kwargs = {})
%convert_element_type_71 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_71, torch.float16), kwargs = {})
%convert_element_type_72 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_45, torch.float16), kwargs = {})
%convert_element_type_73 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_22, torch.float16), kwargs = {})
%squeeze_52 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_72, 3), kwargs = {})
%squeeze_53 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_52, 2), kwargs = {})
%squeeze_54 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_73, 3), kwargs = {})
%squeeze_55 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_54, 2), kwargs = {})
%detach_32 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_53,), kwargs = {})
%detach_33 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_55,), kwargs = {})
%silu_16 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_71,), kwargs = {})
%_param_constant143 : [#users=1] = get_attr[target=_param_constant143]
%_param_constant144 : [#users=1] = get_attr[target=_param_constant144]
%convolution_14 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_16, %_param_constant143, %_param_constant144, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_17 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant145 : [#users=1] = get_attr[target=_param_constant145]
%t_43 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant145,), kwargs = {})
%_param_constant146 : [#users=1] = get_attr[target=_param_constant146]
%addmm_22 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant146, %silu_17, %t_43), kwargs = {})
%slice_23 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_22, 0, 0, 9223372036854775807), kwargs = {})
%slice_24 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_23, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_96 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_24, 2), kwargs = {})
%unsqueeze_97 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_96, 3), kwargs = {})
%add_72 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_14, %unsqueeze_97), kwargs = {})
%view_112 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_72, [2, 32, 16, 1024]), kwargs = {})
%convert_element_type_74 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_112, torch.float32), kwargs = {})
%var_mean_23 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_74, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_46 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_23, 0), kwargs = {})
%getitem_47 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_23, 1), kwargs = {})
%add_73 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_46, 1e-05), kwargs = {})
%rsqrt_23 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_73,), kwargs = {})
%sub_23 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_112, %getitem_47), kwargs = {})
%mul_52 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_23, %rsqrt_23), kwargs = {})
%view_113 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_52, [2, 512, 32, 32]), kwargs = {})
%_param_constant147 : [#users=1] = get_attr[target=_param_constant147]
%unsqueeze_98 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant147, 0), kwargs = {})
%unsqueeze_99 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_98, 2), kwargs = {})
%unsqueeze_100 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_99, 3), kwargs = {})
%_param_constant148 : [#users=1] = get_attr[target=_param_constant148]
%unsqueeze_101 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant148, 0), kwargs = {})
%unsqueeze_102 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_101, 2), kwargs = {})
%unsqueeze_103 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_102, 3), kwargs = {})
%mul_53 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_113, %unsqueeze_103), kwargs = {})
%add_74 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_53, %unsqueeze_100), kwargs = {})
%convert_element_type_75 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_74, torch.float16), kwargs = {})
%convert_element_type_76 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_47, torch.float16), kwargs = {})
%convert_element_type_77 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_23, torch.float16), kwargs = {})
%squeeze_56 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_76, 3), kwargs = {})
%squeeze_57 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_56, 2), kwargs = {})
%squeeze_58 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_77, 3), kwargs = {})
%squeeze_59 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_58, 2), kwargs = {})
%detach_34 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_57,), kwargs = {})
%detach_35 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_59,), kwargs = {})
%silu_18 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_75,), kwargs = {})
%_param_constant149 : [#users=1] = get_attr[target=_param_constant149]
%_param_constant150 : [#users=1] = get_attr[target=_param_constant150]
%convolution_15 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_18, %_param_constant149, %_param_constant150, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%add_75 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%add_69, %convolution_15), kwargs = {})
%div_6 : [#users=2] = call_function[target=torch.ops.aten.div](args = (%add_75, 1.0), kwargs = {})
%view_114 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%div_6, [2, 32, 16, 1024]), kwargs = {})
%convert_element_type_78 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_114, torch.float32), kwargs = {})
%var_mean_24 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_78, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_48 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_24, 0), kwargs = {})
%getitem_49 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_24, 1), kwargs = {})
%add_76 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_48, 1e-06), kwargs = {})
%rsqrt_24 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_76,), kwargs = {})
%sub_24 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_114, %getitem_49), kwargs = {})
%mul_54 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_24, %rsqrt_24), kwargs = {})
%view_115 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_54, [2, 512, 32, 32]), kwargs = {})
%_param_constant151 : [#users=1] = get_attr[target=_param_constant151]
%unsqueeze_104 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant151, 0), kwargs = {})
%unsqueeze_105 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_104, 2), kwargs = {})
%unsqueeze_106 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_105, 3), kwargs = {})
%_param_constant152 : [#users=1] = get_attr[target=_param_constant152]
%unsqueeze_107 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant152, 0), kwargs = {})
%unsqueeze_108 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_107, 2), kwargs = {})
%unsqueeze_109 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_108, 3), kwargs = {})
%mul_55 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_115, %unsqueeze_109), kwargs = {})
%add_77 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_55, %unsqueeze_106), kwargs = {})
%convert_element_type_79 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_77, torch.float16), kwargs = {})
%convert_element_type_80 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_49, torch.float16), kwargs = {})
%convert_element_type_81 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_24, torch.float16), kwargs = {})
%squeeze_60 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_80, 3), kwargs = {})
%squeeze_61 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_60, 2), kwargs = {})
%squeeze_62 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_81, 3), kwargs = {})
%squeeze_63 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_62, 2), kwargs = {})
%detach_36 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_61,), kwargs = {})
%detach_37 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_63,), kwargs = {})
%permute_30 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%convert_element_type_79, [0, 2, 3, 1]), kwargs = {})
%view_116 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%permute_30, [2, 1024, 512]), kwargs = {})
%_param_constant153 : [#users=1] = get_attr[target=_param_constant153]
%t_44 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant153,), kwargs = {})
%expand_7 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%view_116, [2, 1024, 512]), kwargs = {})
%view_117 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_7, [2, 1024, 512]), kwargs = {})
%expand_8 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%t_44, [2, 512, 512]), kwargs = {})
%view_118 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_8, [2, 512, 512]), kwargs = {})
%bmm_9 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%view_117, %view_118), kwargs = {})
%_unsafe_view_45 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%bmm_9, [2, 1024, 512]), kwargs = {})
%_param_constant154 : [#users=1] = get_attr[target=_param_constant154]
%add_78 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%_unsafe_view_45, %_param_constant154), kwargs = {})
%convert_element_type_82 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_78, torch.float32), kwargs = {})
%var_mean_25 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_82, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_50 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_25, 0), kwargs = {})
%getitem_51 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_25, 1), kwargs = {})
%add_79 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_50, 1e-05), kwargs = {})
%rsqrt_25 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_79,), kwargs = {})
%sub_25 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_78, %getitem_51), kwargs = {})
%mul_56 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_25, %rsqrt_25), kwargs = {})
%_param_constant155 : [#users=1] = get_attr[target=_param_constant155]
%mul_57 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_56, %_param_constant155), kwargs = {})
%_param_constant156 : [#users=1] = get_attr[target=_param_constant156]
%add_80 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_57, %_param_constant156), kwargs = {})
%convert_element_type_83 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_80, torch.float16), kwargs = {})
%_param_constant157 : [#users=1] = get_attr[target=_param_constant157]
%t_45 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant157,), kwargs = {})
%view_119 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_83, [2048, 512]), kwargs = {})
%mm_18 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_119, %t_45), kwargs = {})
%_unsafe_view_46 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_18, [2, 1024, 512]), kwargs = {})
%view_120 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_46, [2, 1024, 8, 64]), kwargs = {})
%permute_31 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_120, [0, 2, 1, 3]), kwargs = {})
%clone_27 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_31,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_47 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_27, [16, 1024, 64]), kwargs = {})
%_param_constant158 : [#users=1] = get_attr[target=_param_constant158]
%t_46 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant158,), kwargs = {})
%view_121 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_19 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_121, %t_46), kwargs = {})
%_unsafe_view_48 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_19, [2, 77, 512]), kwargs = {})
%_param_constant159 : [#users=1] = get_attr[target=_param_constant159]
%t_47 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant159,), kwargs = {})
%view_122 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_20 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_122, %t_47), kwargs = {})
%_unsafe_view_49 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_20, [2, 77, 512]), kwargs = {})
%view_123 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_48, [2, 77, 8, 64]), kwargs = {})
%permute_32 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_123, [0, 2, 1, 3]), kwargs = {})
%clone_28 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_32,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_50 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_28, [16, 77, 64]), kwargs = {})
%view_124 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_49, [2, 77, 8, 64]), kwargs = {})
%permute_33 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_124, [0, 2, 1, 3]), kwargs = {})
%clone_29 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_33,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_51 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_29, [16, 77, 64]), kwargs = {})
%empty_6 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 1024, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_6 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_50, -1, -2), kwargs = {})
%baddbmm_6 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_6, %_unsafe_view_47, %transpose_6), kwargs = {beta: 0, alpha: 0.125})
%_softmax_6 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_6, -1, False), kwargs = {})
%detach_38 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_6,), kwargs = {})
%bmm_10 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_6, %_unsafe_view_51), kwargs = {})
%view_125 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_10, [2, 8, 1024, 64]), kwargs = {})
%permute_34 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_125, [0, 2, 1, 3]), kwargs = {})
%clone_30 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_34,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_52 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_30, [2, 1024, 512]), kwargs = {})
%_param_constant160 : [#users=1] = get_attr[target=_param_constant160]
%t_48 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant160,), kwargs = {})
%view_126 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_52, [2048, 512]), kwargs = {})
%_param_constant161 : [#users=1] = get_attr[target=_param_constant161]
%addmm_23 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant161, %view_126, %t_48), kwargs = {})
%view_127 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_23, [2, 1024, 512]), kwargs = {})
%add_81 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_127, %add_78), kwargs = {})
%convert_element_type_84 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_81, torch.float32), kwargs = {})
%var_mean_26 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_84, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_52 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_26, 0), kwargs = {})
%getitem_53 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_26, 1), kwargs = {})
%add_82 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_52, 1e-05), kwargs = {})
%rsqrt_26 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_82,), kwargs = {})
%sub_26 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_81, %getitem_53), kwargs = {})
%mul_58 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_26, %rsqrt_26), kwargs = {})
%_param_constant162 : [#users=1] = get_attr[target=_param_constant162]
%mul_59 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_58, %_param_constant162), kwargs = {})
%_param_constant163 : [#users=1] = get_attr[target=_param_constant163]
%add_83 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_59, %_param_constant163), kwargs = {})
%convert_element_type_85 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_83, torch.float16), kwargs = {})
%_param_constant164 : [#users=1] = get_attr[target=_param_constant164]
%t_49 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant164,), kwargs = {})
%view_128 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_85, [2048, 512]), kwargs = {})
%mm_21 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_128, %t_49), kwargs = {})
%_unsafe_view_53 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_21, [2, 1024, 512]), kwargs = {})
%view_129 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_53, [2, 1024, 8, 64]), kwargs = {})
%permute_35 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_129, [0, 2, 1, 3]), kwargs = {})
%clone_31 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_35,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_54 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_31, [16, 1024, 64]), kwargs = {})
%_param_constant165 : [#users=1] = get_attr[target=_param_constant165]
%t_50 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant165,), kwargs = {})
%view_130 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_22 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_130, %t_50), kwargs = {})
%_unsafe_view_55 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_22, [2, 77, 512]), kwargs = {})
%_param_constant166 : [#users=1] = get_attr[target=_param_constant166]
%t_51 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant166,), kwargs = {})
%view_131 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_23 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_131, %t_51), kwargs = {})
%_unsafe_view_56 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_23, [2, 77, 512]), kwargs = {})
%view_132 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_55, [2, 77, 8, 64]), kwargs = {})
%permute_36 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_132, [0, 2, 1, 3]), kwargs = {})
%clone_32 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_36,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_57 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_32, [16, 77, 64]), kwargs = {})
%view_133 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_56, [2, 77, 8, 64]), kwargs = {})
%permute_37 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_133, [0, 2, 1, 3]), kwargs = {})
%clone_33 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_37,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_58 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_33, [16, 77, 64]), kwargs = {})
%empty_7 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 1024, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_7 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_57, -1, -2), kwargs = {})
%baddbmm_7 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_7, %_unsafe_view_54, %transpose_7), kwargs = {beta: 0, alpha: 0.125})
%_softmax_7 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_7, -1, False), kwargs = {})
%detach_39 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_7,), kwargs = {})
%bmm_11 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_7, %_unsafe_view_58), kwargs = {})
%view_134 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_11, [2, 8, 1024, 64]), kwargs = {})
%permute_38 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_134, [0, 2, 1, 3]), kwargs = {})
%clone_34 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_38,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_59 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_34, [2, 1024, 512]), kwargs = {})
%_param_constant167 : [#users=1] = get_attr[target=_param_constant167]
%t_52 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant167,), kwargs = {})
%view_135 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_59, [2048, 512]), kwargs = {})
%_param_constant168 : [#users=1] = get_attr[target=_param_constant168]
%addmm_24 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant168, %view_135, %t_52), kwargs = {})
%view_136 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_24, [2, 1024, 512]), kwargs = {})
%add_84 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_136, %add_81), kwargs = {})
%convert_element_type_86 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_84, torch.float32), kwargs = {})
%var_mean_27 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_86, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_54 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_27, 0), kwargs = {})
%getitem_55 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_27, 1), kwargs = {})
%add_85 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_54, 1e-05), kwargs = {})
%rsqrt_27 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_85,), kwargs = {})
%sub_27 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_84, %getitem_55), kwargs = {})
%mul_60 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_27, %rsqrt_27), kwargs = {})
%_param_constant169 : [#users=1] = get_attr[target=_param_constant169]
%mul_61 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_60, %_param_constant169), kwargs = {})
%_param_constant170 : [#users=1] = get_attr[target=_param_constant170]
%add_86 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_61, %_param_constant170), kwargs = {})
%convert_element_type_87 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_86, torch.float16), kwargs = {})
%_param_constant171 : [#users=1] = get_attr[target=_param_constant171]
%t_53 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant171,), kwargs = {})
%view_137 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_87, [2048, 512]), kwargs = {})
%_param_constant172 : [#users=1] = get_attr[target=_param_constant172]
%addmm_25 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant172, %view_137, %t_53), kwargs = {})
%view_138 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%addmm_25, [2, 1024, 4096]), kwargs = {})
%slice_25 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_138, -1, 0, 2048), kwargs = {})
%slice_26 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_138, -1, 2048, 4096), kwargs = {})
%gelu_3 : [#users=1] = call_function[target=torch.ops.aten.gelu](args = (%slice_26,), kwargs = {})
%mul_62 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%slice_25, %gelu_3), kwargs = {})
%_param_constant173 : [#users=1] = get_attr[target=_param_constant173]
%t_54 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant173,), kwargs = {})
%view_139 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_62, [2048, 2048]), kwargs = {})
%_param_constant174 : [#users=1] = get_attr[target=_param_constant174]
%addmm_26 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant174, %view_139, %t_54), kwargs = {})
%view_140 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_26, [2, 1024, 512]), kwargs = {})
%add_87 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%view_140, %add_84), kwargs = {})
%_param_constant175 : [#users=1] = get_attr[target=_param_constant175]
%t_55 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant175,), kwargs = {})
%view_141 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%add_87, [2048, 512]), kwargs = {})
%_param_constant176 : [#users=1] = get_attr[target=_param_constant176]
%addmm_27 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant176, %view_141, %t_55), kwargs = {})
%view_142 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_27, [2, 1024, 512]), kwargs = {})
%view_143 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%view_142, [2, 32, 32, 512]), kwargs = {})
%permute_39 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_143, [0, 3, 1, 2]), kwargs = {})
%clone_35 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_39,), kwargs = {memory_format: torch.contiguous_format})
%add_88 : [#users=2] = call_function[target=torch.ops.aten.add](args = (%clone_35, %div_6), kwargs = {})
%_param_constant177 : [#users=1] = get_attr[target=_param_constant177]
%_param_constant178 : [#users=1] = get_attr[target=_param_constant178]
%convolution_16 : [#users=3] = call_function[target=torch.ops.aten.convolution](args = (%add_88, %_param_constant177, %_param_constant178, [2, 2], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%view_144 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%convolution_16, [2, 32, 16, 256]), kwargs = {})
%convert_element_type_88 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_144, torch.float32), kwargs = {})
%var_mean_28 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_88, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_56 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_28, 0), kwargs = {})
%getitem_57 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_28, 1), kwargs = {})
%add_89 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_56, 1e-05), kwargs = {})
%rsqrt_28 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_89,), kwargs = {})
%sub_28 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_144, %getitem_57), kwargs = {})
%mul_63 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_28, %rsqrt_28), kwargs = {})
%view_145 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_63, [2, 512, 16, 16]), kwargs = {})
%_param_constant179 : [#users=1] = get_attr[target=_param_constant179]
%unsqueeze_110 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant179, 0), kwargs = {})
%unsqueeze_111 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_110, 2), kwargs = {})
%unsqueeze_112 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_111, 3), kwargs = {})
%_param_constant180 : [#users=1] = get_attr[target=_param_constant180]
%unsqueeze_113 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant180, 0), kwargs = {})
%unsqueeze_114 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_113, 2), kwargs = {})
%unsqueeze_115 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_114, 3), kwargs = {})
%mul_64 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_145, %unsqueeze_115), kwargs = {})
%add_90 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_64, %unsqueeze_112), kwargs = {})
%convert_element_type_89 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_90, torch.float16), kwargs = {})
%convert_element_type_90 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_57, torch.float16), kwargs = {})
%convert_element_type_91 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_28, torch.float16), kwargs = {})
%squeeze_64 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_90, 3), kwargs = {})
%squeeze_65 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_64, 2), kwargs = {})
%squeeze_66 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_91, 3), kwargs = {})
%squeeze_67 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_66, 2), kwargs = {})
%detach_40 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_65,), kwargs = {})
%detach_41 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_67,), kwargs = {})
%silu_19 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_89,), kwargs = {})
%_param_constant181 : [#users=1] = get_attr[target=_param_constant181]
%_param_constant182 : [#users=1] = get_attr[target=_param_constant182]
%convolution_17 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_19, %_param_constant181, %_param_constant182, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_20 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant183 : [#users=1] = get_attr[target=_param_constant183]
%t_56 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant183,), kwargs = {})
%_param_constant184 : [#users=1] = get_attr[target=_param_constant184]
%addmm_28 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant184, %silu_20, %t_56), kwargs = {})
%slice_27 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_28, 0, 0, 9223372036854775807), kwargs = {})
%slice_28 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_27, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_116 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_28, 2), kwargs = {})
%unsqueeze_117 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_116, 3), kwargs = {})
%add_91 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_17, %unsqueeze_117), kwargs = {})
%view_146 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_91, [2, 32, 32, 256]), kwargs = {})
%convert_element_type_92 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_146, torch.float32), kwargs = {})
%var_mean_29 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_92, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_58 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_29, 0), kwargs = {})
%getitem_59 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_29, 1), kwargs = {})
%add_92 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_58, 1e-05), kwargs = {})
%rsqrt_29 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_92,), kwargs = {})
%sub_29 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_146, %getitem_59), kwargs = {})
%mul_65 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_29, %rsqrt_29), kwargs = {})
%view_147 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_65, [2, 1024, 16, 16]), kwargs = {})
%_param_constant185 : [#users=1] = get_attr[target=_param_constant185]
%unsqueeze_118 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant185, 0), kwargs = {})
%unsqueeze_119 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_118, 2), kwargs = {})
%unsqueeze_120 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_119, 3), kwargs = {})
%_param_constant186 : [#users=1] = get_attr[target=_param_constant186]
%unsqueeze_121 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant186, 0), kwargs = {})
%unsqueeze_122 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_121, 2), kwargs = {})
%unsqueeze_123 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_122, 3), kwargs = {})
%mul_66 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_147, %unsqueeze_123), kwargs = {})
%add_93 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_66, %unsqueeze_120), kwargs = {})
%convert_element_type_93 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_93, torch.float16), kwargs = {})
%convert_element_type_94 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_59, torch.float16), kwargs = {})
%convert_element_type_95 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_29, torch.float16), kwargs = {})
%squeeze_68 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_94, 3), kwargs = {})
%squeeze_69 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_68, 2), kwargs = {})
%squeeze_70 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_95, 3), kwargs = {})
%squeeze_71 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_70, 2), kwargs = {})
%detach_42 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_69,), kwargs = {})
%detach_43 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_71,), kwargs = {})
%silu_21 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_93,), kwargs = {})
%_param_constant187 : [#users=1] = get_attr[target=_param_constant187]
%_param_constant188 : [#users=1] = get_attr[target=_param_constant188]
%convolution_18 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_21, %_param_constant187, %_param_constant188, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%_param_constant189 : [#users=1] = get_attr[target=_param_constant189]
%_param_constant190 : [#users=1] = get_attr[target=_param_constant190]
%convolution_19 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%convolution_16, %_param_constant189, %_param_constant190, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
%add_94 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_19, %convolution_18), kwargs = {})
%div_7 : [#users=2] = call_function[target=torch.ops.aten.div](args = (%add_94, 1.0), kwargs = {})
%view_148 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%div_7, [2, 32, 32, 256]), kwargs = {})
%convert_element_type_96 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_148, torch.float32), kwargs = {})
%var_mean_30 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_96, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_60 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_30, 0), kwargs = {})
%getitem_61 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_30, 1), kwargs = {})
%add_95 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_60, 1e-06), kwargs = {})
%rsqrt_30 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_95,), kwargs = {})
%sub_30 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_148, %getitem_61), kwargs = {})
%mul_67 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_30, %rsqrt_30), kwargs = {})
%view_149 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_67, [2, 1024, 16, 16]), kwargs = {})
%_param_constant191 : [#users=1] = get_attr[target=_param_constant191]
%unsqueeze_124 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant191, 0), kwargs = {})
%unsqueeze_125 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_124, 2), kwargs = {})
%unsqueeze_126 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_125, 3), kwargs = {})
%_param_constant192 : [#users=1] = get_attr[target=_param_constant192]
%unsqueeze_127 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant192, 0), kwargs = {})
%unsqueeze_128 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_127, 2), kwargs = {})
%unsqueeze_129 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_128, 3), kwargs = {})
%mul_68 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_149, %unsqueeze_129), kwargs = {})
%add_96 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_68, %unsqueeze_126), kwargs = {})
%convert_element_type_97 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_96, torch.float16), kwargs = {})
%convert_element_type_98 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_61, torch.float16), kwargs = {})
%convert_element_type_99 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_30, torch.float16), kwargs = {})
%squeeze_72 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_98, 3), kwargs = {})
%squeeze_73 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_72, 2), kwargs = {})
%squeeze_74 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_99, 3), kwargs = {})
%squeeze_75 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_74, 2), kwargs = {})
%detach_44 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_73,), kwargs = {})
%detach_45 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_75,), kwargs = {})
%permute_40 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%convert_element_type_97, [0, 2, 3, 1]), kwargs = {})
%view_150 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%permute_40, [2, 256, 1024]), kwargs = {})
%_param_constant193 : [#users=1] = get_attr[target=_param_constant193]
%t_57 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant193,), kwargs = {})
%expand_9 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%view_150, [2, 256, 1024]), kwargs = {})
%view_151 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_9, [2, 256, 1024]), kwargs = {})
%expand_10 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%t_57, [2, 1024, 1024]), kwargs = {})
%view_152 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_10, [2, 1024, 1024]), kwargs = {})
%bmm_12 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%view_151, %view_152), kwargs = {})
%_unsafe_view_60 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%bmm_12, [2, 256, 1024]), kwargs = {})
%_param_constant194 : [#users=1] = get_attr[target=_param_constant194]
%add_97 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%_unsafe_view_60, %_param_constant194), kwargs = {})
%convert_element_type_100 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_97, torch.float32), kwargs = {})
%var_mean_31 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_100, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_62 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_31, 0), kwargs = {})
%getitem_63 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_31, 1), kwargs = {})
%add_98 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_62, 1e-05), kwargs = {})
%rsqrt_31 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_98,), kwargs = {})
%sub_31 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_97, %getitem_63), kwargs = {})
%mul_69 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_31, %rsqrt_31), kwargs = {})
%_param_constant195 : [#users=1] = get_attr[target=_param_constant195]
%mul_70 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_69, %_param_constant195), kwargs = {})
%_param_constant196 : [#users=1] = get_attr[target=_param_constant196]
%add_99 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_70, %_param_constant196), kwargs = {})
%convert_element_type_101 : [#users=3] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_99, torch.float16), kwargs = {})
%_param_constant197 : [#users=1] = get_attr[target=_param_constant197]
%t_58 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant197,), kwargs = {})
%view_153 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_101, [512, 1024]), kwargs = {})
%mm_24 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_153, %t_58), kwargs = {})
%_unsafe_view_61 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_24, [2, 256, 1024]), kwargs = {})
%view_154 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_61, [2, 256, 8, 128]), kwargs = {})
%permute_41 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_154, [0, 2, 1, 3]), kwargs = {})
%clone_36 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_41,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_62 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_36, [16, 256, 128]), kwargs = {})
%_param_constant198 : [#users=1] = get_attr[target=_param_constant198]
%t_59 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant198,), kwargs = {})
%view_155 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_101, [512, 1024]), kwargs = {})
%mm_25 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_155, %t_59), kwargs = {})
%_unsafe_view_63 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_25, [2, 256, 1024]), kwargs = {})
%_param_constant199 : [#users=1] = get_attr[target=_param_constant199]
%t_60 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant199,), kwargs = {})
%view_156 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_101, [512, 1024]), kwargs = {})
%mm_26 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_156, %t_60), kwargs = {})
%_unsafe_view_64 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_26, [2, 256, 1024]), kwargs = {})
%view_157 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_63, [2, 256, 8, 128]), kwargs = {})
%permute_42 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_157, [0, 2, 1, 3]), kwargs = {})
%clone_37 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_42,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_65 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_37, [16, 256, 128]), kwargs = {})
%view_158 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_64, [2, 256, 8, 128]), kwargs = {})
%permute_43 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_158, [0, 2, 1, 3]), kwargs = {})
%clone_38 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_43,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_66 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_38, [16, 256, 128]), kwargs = {})
%empty_8 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 256, 256],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_8 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_65, -1, -2), kwargs = {})
%baddbmm_8 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_8, %_unsafe_view_62, %transpose_8), kwargs = {beta: 0, alpha: 0.08838834764831845})
%_softmax_8 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_8, -1, False), kwargs = {})
%detach_46 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_8,), kwargs = {})
%bmm_13 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_8, %_unsafe_view_66), kwargs = {})
%view_159 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_13, [2, 8, 256, 128]), kwargs = {})
%permute_44 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_159, [0, 2, 1, 3]), kwargs = {})
%clone_39 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_44,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_67 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_39, [2, 256, 1024]), kwargs = {})
%_param_constant200 : [#users=1] = get_attr[target=_param_constant200]
%t_61 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant200,), kwargs = {})
%view_160 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_67, [512, 1024]), kwargs = {})
%_param_constant201 : [#users=1] = get_attr[target=_param_constant201]
%addmm_29 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant201, %view_160, %t_61), kwargs = {})
%view_161 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_29, [2, 256, 1024]), kwargs = {})
%add_100 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_161, %add_97), kwargs = {})
%convert_element_type_102 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_100, torch.float32), kwargs = {})
%var_mean_32 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_102, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_64 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_32, 0), kwargs = {})
%getitem_65 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_32, 1), kwargs = {})
%add_101 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_64, 1e-05), kwargs = {})
%rsqrt_32 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_101,), kwargs = {})
%sub_32 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_100, %getitem_65), kwargs = {})
%mul_71 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_32, %rsqrt_32), kwargs = {})
%_param_constant202 : [#users=1] = get_attr[target=_param_constant202]
%mul_72 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_71, %_param_constant202), kwargs = {})
%_param_constant203 : [#users=1] = get_attr[target=_param_constant203]
%add_102 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_72, %_param_constant203), kwargs = {})
%convert_element_type_103 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_102, torch.float16), kwargs = {})
%_param_constant204 : [#users=1] = get_attr[target=_param_constant204]
%t_62 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant204,), kwargs = {})
%view_162 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_103, [512, 1024]), kwargs = {})
%mm_27 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_162, %t_62), kwargs = {})
%_unsafe_view_68 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_27, [2, 256, 1024]), kwargs = {})
%view_163 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_68, [2, 256, 8, 128]), kwargs = {})
%permute_45 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_163, [0, 2, 1, 3]), kwargs = {})
%clone_40 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_45,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_69 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_40, [16, 256, 128]), kwargs = {})
%_param_constant205 : [#users=1] = get_attr[target=_param_constant205]
%t_63 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant205,), kwargs = {})
%view_164 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_28 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_164, %t_63), kwargs = {})
%_unsafe_view_70 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_28, [2, 77, 1024]), kwargs = {})
%_param_constant206 : [#users=1] = get_attr[target=_param_constant206]
%t_64 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant206,), kwargs = {})
%view_165 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_29 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_165, %t_64), kwargs = {})
%_unsafe_view_71 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_29, [2, 77, 1024]), kwargs = {})
%view_166 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_70, [2, 77, 8, 128]), kwargs = {})
%permute_46 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_166, [0, 2, 1, 3]), kwargs = {})
%clone_41 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_46,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_72 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_41, [16, 77, 128]), kwargs = {})
%view_167 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_71, [2, 77, 8, 128]), kwargs = {})
%permute_47 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_167, [0, 2, 1, 3]), kwargs = {})
%clone_42 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_47,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_73 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_42, [16, 77, 128]), kwargs = {})
%empty_9 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 256, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_9 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_72, -1, -2), kwargs = {})
%baddbmm_9 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_9, %_unsafe_view_69, %transpose_9), kwargs = {beta: 0, alpha: 0.08838834764831845})
%_softmax_9 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_9, -1, False), kwargs = {})
%detach_47 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_9,), kwargs = {})
%bmm_14 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_9, %_unsafe_view_73), kwargs = {})
%view_168 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_14, [2, 8, 256, 128]), kwargs = {})
%permute_48 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_168, [0, 2, 1, 3]), kwargs = {})
%clone_43 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_48,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_74 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_43, [2, 256, 1024]), kwargs = {})
%_param_constant207 : [#users=1] = get_attr[target=_param_constant207]
%t_65 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant207,), kwargs = {})
%view_169 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_74, [512, 1024]), kwargs = {})
%_param_constant208 : [#users=1] = get_attr[target=_param_constant208]
%addmm_30 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant208, %view_169, %t_65), kwargs = {})
%view_170 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_30, [2, 256, 1024]), kwargs = {})
%add_103 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_170, %add_100), kwargs = {})
%convert_element_type_104 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_103, torch.float32), kwargs = {})
%var_mean_33 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_104, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_66 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_33, 0), kwargs = {})
%getitem_67 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_33, 1), kwargs = {})
%add_104 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_66, 1e-05), kwargs = {})
%rsqrt_33 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_104,), kwargs = {})
%sub_33 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_103, %getitem_67), kwargs = {})
%mul_73 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_33, %rsqrt_33), kwargs = {})
%_param_constant209 : [#users=1] = get_attr[target=_param_constant209]
%mul_74 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_73, %_param_constant209), kwargs = {})
%_param_constant210 : [#users=1] = get_attr[target=_param_constant210]
%add_105 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_74, %_param_constant210), kwargs = {})
%convert_element_type_105 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_105, torch.float16), kwargs = {})
%_param_constant211 : [#users=1] = get_attr[target=_param_constant211]
%t_66 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant211,), kwargs = {})
%view_171 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_105, [512, 1024]), kwargs = {})
%_param_constant212 : [#users=1] = get_attr[target=_param_constant212]
%addmm_31 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant212, %view_171, %t_66), kwargs = {})
%view_172 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%addmm_31, [2, 256, 8192]), kwargs = {})
%slice_29 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_172, -1, 0, 4096), kwargs = {})
%slice_30 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_172, -1, 4096, 8192), kwargs = {})
%gelu_4 : [#users=1] = call_function[target=torch.ops.aten.gelu](args = (%slice_30,), kwargs = {})
%mul_75 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%slice_29, %gelu_4), kwargs = {})
%_param_constant213 : [#users=1] = get_attr[target=_param_constant213]
%t_67 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant213,), kwargs = {})
%view_173 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_75, [512, 4096]), kwargs = {})
%_param_constant214 : [#users=1] = get_attr[target=_param_constant214]
%addmm_32 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant214, %view_173, %t_67), kwargs = {})
%view_174 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_32, [2, 256, 1024]), kwargs = {})
%add_106 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%view_174, %add_103), kwargs = {})
%_param_constant215 : [#users=1] = get_attr[target=_param_constant215]
%t_68 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant215,), kwargs = {})
%view_175 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%add_106, [512, 1024]), kwargs = {})
%_param_constant216 : [#users=1] = get_attr[target=_param_constant216]
%addmm_33 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant216, %view_175, %t_68), kwargs = {})
%view_176 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_33, [2, 256, 1024]), kwargs = {})
%view_177 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%view_176, [2, 16, 16, 1024]), kwargs = {})
%permute_49 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_177, [0, 3, 1, 2]), kwargs = {})
%clone_44 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_49,), kwargs = {memory_format: torch.contiguous_format})
%add_107 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%clone_44, %div_7), kwargs = {})
%view_178 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_107, [2, 32, 32, 256]), kwargs = {})
%convert_element_type_106 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_178, torch.float32), kwargs = {})
%var_mean_34 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_106, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_68 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_34, 0), kwargs = {})
%getitem_69 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_34, 1), kwargs = {})
%add_108 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_68, 1e-05), kwargs = {})
%rsqrt_34 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_108,), kwargs = {})
%sub_34 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_178, %getitem_69), kwargs = {})
%mul_76 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_34, %rsqrt_34), kwargs = {})
%view_179 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_76, [2, 1024, 16, 16]), kwargs = {})
%_param_constant217 : [#users=1] = get_attr[target=_param_constant217]
%unsqueeze_130 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant217, 0), kwargs = {})
%unsqueeze_131 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_130, 2), kwargs = {})
%unsqueeze_132 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_131, 3), kwargs = {})
%_param_constant218 : [#users=1] = get_attr[target=_param_constant218]
%unsqueeze_133 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant218, 0), kwargs = {})
%unsqueeze_134 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_133, 2), kwargs = {})
%unsqueeze_135 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_134, 3), kwargs = {})
%mul_77 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_179, %unsqueeze_135), kwargs = {})
%add_109 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_77, %unsqueeze_132), kwargs = {})
%convert_element_type_107 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_109, torch.float16), kwargs = {})
%convert_element_type_108 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_69, torch.float16), kwargs = {})
%convert_element_type_109 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_34, torch.float16), kwargs = {})
%squeeze_76 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_108, 3), kwargs = {})
%squeeze_77 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_76, 2), kwargs = {})
%squeeze_78 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_109, 3), kwargs = {})
%squeeze_79 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_78, 2), kwargs = {})
%detach_48 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_77,), kwargs = {})
%detach_49 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_79,), kwargs = {})
%silu_22 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_107,), kwargs = {})
%_param_constant219 : [#users=1] = get_attr[target=_param_constant219]
%_param_constant220 : [#users=1] = get_attr[target=_param_constant220]
%convolution_20 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_22, %_param_constant219, %_param_constant220, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_23 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant221 : [#users=1] = get_attr[target=_param_constant221]
%t_69 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant221,), kwargs = {})
%_param_constant222 : [#users=1] = get_attr[target=_param_constant222]
%addmm_34 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant222, %silu_23, %t_69), kwargs = {})
%slice_31 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_34, 0, 0, 9223372036854775807), kwargs = {})
%slice_32 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_31, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_136 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_32, 2), kwargs = {})
%unsqueeze_137 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_136, 3), kwargs = {})
%add_110 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_20, %unsqueeze_137), kwargs = {})
%view_180 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_110, [2, 32, 32, 256]), kwargs = {})
%convert_element_type_110 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_180, torch.float32), kwargs = {})
%var_mean_35 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_110, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_70 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_35, 0), kwargs = {})
%getitem_71 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_35, 1), kwargs = {})
%add_111 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_70, 1e-05), kwargs = {})
%rsqrt_35 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_111,), kwargs = {})
%sub_35 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_180, %getitem_71), kwargs = {})
%mul_78 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_35, %rsqrt_35), kwargs = {})
%view_181 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_78, [2, 1024, 16, 16]), kwargs = {})
%_param_constant223 : [#users=1] = get_attr[target=_param_constant223]
%unsqueeze_138 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant223, 0), kwargs = {})
%unsqueeze_139 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_138, 2), kwargs = {})
%unsqueeze_140 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_139, 3), kwargs = {})
%_param_constant224 : [#users=1] = get_attr[target=_param_constant224]
%unsqueeze_141 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant224, 0), kwargs = {})
%unsqueeze_142 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_141, 2), kwargs = {})
%unsqueeze_143 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_142, 3), kwargs = {})
%mul_79 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_181, %unsqueeze_143), kwargs = {})
%add_112 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_79, %unsqueeze_140), kwargs = {})
%convert_element_type_111 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_112, torch.float16), kwargs = {})
%convert_element_type_112 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_71, torch.float16), kwargs = {})
%convert_element_type_113 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_35, torch.float16), kwargs = {})
%squeeze_80 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_112, 3), kwargs = {})
%squeeze_81 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_80, 2), kwargs = {})
%squeeze_82 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_113, 3), kwargs = {})
%squeeze_83 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_82, 2), kwargs = {})
%detach_50 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_81,), kwargs = {})
%detach_51 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_83,), kwargs = {})
%silu_24 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_111,), kwargs = {})
%_param_constant225 : [#users=1] = get_attr[target=_param_constant225]
%_param_constant226 : [#users=1] = get_attr[target=_param_constant226]
%convolution_21 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_24, %_param_constant225, %_param_constant226, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%add_113 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%add_107, %convolution_21), kwargs = {})
%div_8 : [#users=2] = call_function[target=torch.ops.aten.div](args = (%add_113, 1.0), kwargs = {})
%view_182 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%div_8, [2, 32, 32, 256]), kwargs = {})
%convert_element_type_114 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_182, torch.float32), kwargs = {})
%var_mean_36 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_114, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_72 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_36, 0), kwargs = {})
%getitem_73 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_36, 1), kwargs = {})
%add_114 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_72, 1e-06), kwargs = {})
%rsqrt_36 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_114,), kwargs = {})
%sub_36 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_182, %getitem_73), kwargs = {})
%mul_80 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_36, %rsqrt_36), kwargs = {})
%view_183 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_80, [2, 1024, 16, 16]), kwargs = {})
%_param_constant227 : [#users=1] = get_attr[target=_param_constant227]
%unsqueeze_144 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant227, 0), kwargs = {})
%unsqueeze_145 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_144, 2), kwargs = {})
%unsqueeze_146 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_145, 3), kwargs = {})
%_param_constant228 : [#users=1] = get_attr[target=_param_constant228]
%unsqueeze_147 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant228, 0), kwargs = {})
%unsqueeze_148 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_147, 2), kwargs = {})
%unsqueeze_149 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_148, 3), kwargs = {})
%mul_81 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_183, %unsqueeze_149), kwargs = {})
%add_115 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_81, %unsqueeze_146), kwargs = {})
%convert_element_type_115 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_115, torch.float16), kwargs = {})
%convert_element_type_116 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_73, torch.float16), kwargs = {})
%convert_element_type_117 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_36, torch.float16), kwargs = {})
%squeeze_84 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_116, 3), kwargs = {})
%squeeze_85 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_84, 2), kwargs = {})
%squeeze_86 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_117, 3), kwargs = {})
%squeeze_87 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_86, 2), kwargs = {})
%detach_52 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_85,), kwargs = {})
%detach_53 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_87,), kwargs = {})
%permute_50 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%convert_element_type_115, [0, 2, 3, 1]), kwargs = {})
%view_184 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%permute_50, [2, 256, 1024]), kwargs = {})
%_param_constant229 : [#users=1] = get_attr[target=_param_constant229]
%t_70 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant229,), kwargs = {})
%expand_11 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%view_184, [2, 256, 1024]), kwargs = {})
%view_185 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_11, [2, 256, 1024]), kwargs = {})
%expand_12 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%t_70, [2, 1024, 1024]), kwargs = {})
%view_186 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_12, [2, 1024, 1024]), kwargs = {})
%bmm_15 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%view_185, %view_186), kwargs = {})
%_unsafe_view_75 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%bmm_15, [2, 256, 1024]), kwargs = {})
%_param_constant230 : [#users=1] = get_attr[target=_param_constant230]
%add_116 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%_unsafe_view_75, %_param_constant230), kwargs = {})
%convert_element_type_118 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_116, torch.float32), kwargs = {})
%var_mean_37 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_118, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_74 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_37, 0), kwargs = {})
%getitem_75 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_37, 1), kwargs = {})
%add_117 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_74, 1e-05), kwargs = {})
%rsqrt_37 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_117,), kwargs = {})
%sub_37 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_116, %getitem_75), kwargs = {})
%mul_82 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_37, %rsqrt_37), kwargs = {})
%_param_constant231 : [#users=1] = get_attr[target=_param_constant231]
%mul_83 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_82, %_param_constant231), kwargs = {})
%_param_constant232 : [#users=1] = get_attr[target=_param_constant232]
%add_118 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_83, %_param_constant232), kwargs = {})
%convert_element_type_119 : [#users=3] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_118, torch.float16), kwargs = {})
%_param_constant233 : [#users=1] = get_attr[target=_param_constant233]
%t_71 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant233,), kwargs = {})
%view_187 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_119, [512, 1024]), kwargs = {})
%mm_30 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_187, %t_71), kwargs = {})
%_unsafe_view_76 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_30, [2, 256, 1024]), kwargs = {})
%view_188 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_76, [2, 256, 8, 128]), kwargs = {})
%permute_51 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_188, [0, 2, 1, 3]), kwargs = {})
%clone_45 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_51,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_77 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_45, [16, 256, 128]), kwargs = {})
%_param_constant234 : [#users=1] = get_attr[target=_param_constant234]
%t_72 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant234,), kwargs = {})
%view_189 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_119, [512, 1024]), kwargs = {})
%mm_31 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_189, %t_72), kwargs = {})
%_unsafe_view_78 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_31, [2, 256, 1024]), kwargs = {})
%_param_constant235 : [#users=1] = get_attr[target=_param_constant235]
%t_73 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant235,), kwargs = {})
%view_190 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_119, [512, 1024]), kwargs = {})
%mm_32 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_190, %t_73), kwargs = {})
%_unsafe_view_79 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_32, [2, 256, 1024]), kwargs = {})
%view_191 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_78, [2, 256, 8, 128]), kwargs = {})
%permute_52 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_191, [0, 2, 1, 3]), kwargs = {})
%clone_46 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_52,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_80 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_46, [16, 256, 128]), kwargs = {})
%view_192 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_79, [2, 256, 8, 128]), kwargs = {})
%permute_53 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_192, [0, 2, 1, 3]), kwargs = {})
%clone_47 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_53,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_81 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_47, [16, 256, 128]), kwargs = {})
%empty_10 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 256, 256],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_10 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_80, -1, -2), kwargs = {})
%baddbmm_10 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_10, %_unsafe_view_77, %transpose_10), kwargs = {beta: 0, alpha: 0.08838834764831845})
%_softmax_10 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_10, -1, False), kwargs = {})
%detach_54 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_10,), kwargs = {})
%bmm_16 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_10, %_unsafe_view_81), kwargs = {})
%view_193 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_16, [2, 8, 256, 128]), kwargs = {})
%permute_54 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_193, [0, 2, 1, 3]), kwargs = {})
%clone_48 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_54,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_82 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_48, [2, 256, 1024]), kwargs = {})
%_param_constant236 : [#users=1] = get_attr[target=_param_constant236]
%t_74 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant236,), kwargs = {})
%view_194 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_82, [512, 1024]), kwargs = {})
%_param_constant237 : [#users=1] = get_attr[target=_param_constant237]
%addmm_35 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant237, %view_194, %t_74), kwargs = {})
%view_195 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_35, [2, 256, 1024]), kwargs = {})
%add_119 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_195, %add_116), kwargs = {})
%convert_element_type_120 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_119, torch.float32), kwargs = {})
%var_mean_38 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_120, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_76 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_38, 0), kwargs = {})
%getitem_77 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_38, 1), kwargs = {})
%add_120 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_76, 1e-05), kwargs = {})
%rsqrt_38 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_120,), kwargs = {})
%sub_38 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_119, %getitem_77), kwargs = {})
%mul_84 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_38, %rsqrt_38), kwargs = {})
%_param_constant238 : [#users=1] = get_attr[target=_param_constant238]
%mul_85 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_84, %_param_constant238), kwargs = {})
%_param_constant239 : [#users=1] = get_attr[target=_param_constant239]
%add_121 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_85, %_param_constant239), kwargs = {})
%convert_element_type_121 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_121, torch.float16), kwargs = {})
%_param_constant240 : [#users=1] = get_attr[target=_param_constant240]
%t_75 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant240,), kwargs = {})
%view_196 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_121, [512, 1024]), kwargs = {})
%mm_33 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_196, %t_75), kwargs = {})
%_unsafe_view_83 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_33, [2, 256, 1024]), kwargs = {})
%view_197 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_83, [2, 256, 8, 128]), kwargs = {})
%permute_55 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_197, [0, 2, 1, 3]), kwargs = {})
%clone_49 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_55,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_84 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_49, [16, 256, 128]), kwargs = {})
%_param_constant241 : [#users=1] = get_attr[target=_param_constant241]
%t_76 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant241,), kwargs = {})
%view_198 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_34 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_198, %t_76), kwargs = {})
%_unsafe_view_85 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_34, [2, 77, 1024]), kwargs = {})
%_param_constant242 : [#users=1] = get_attr[target=_param_constant242]
%t_77 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant242,), kwargs = {})
%view_199 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_35 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_199, %t_77), kwargs = {})
%_unsafe_view_86 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_35, [2, 77, 1024]), kwargs = {})
%view_200 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_85, [2, 77, 8, 128]), kwargs = {})
%permute_56 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_200, [0, 2, 1, 3]), kwargs = {})
%clone_50 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_56,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_87 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_50, [16, 77, 128]), kwargs = {})
%view_201 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_86, [2, 77, 8, 128]), kwargs = {})
%permute_57 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_201, [0, 2, 1, 3]), kwargs = {})
%clone_51 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_57,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_88 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_51, [16, 77, 128]), kwargs = {})
%empty_11 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 256, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_11 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_87, -1, -2), kwargs = {})
%baddbmm_11 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_11, %_unsafe_view_84, %transpose_11), kwargs = {beta: 0, alpha: 0.08838834764831845})
%_softmax_11 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_11, -1, False), kwargs = {})
%detach_55 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_11,), kwargs = {})
%bmm_17 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_11, %_unsafe_view_88), kwargs = {})
%view_202 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_17, [2, 8, 256, 128]), kwargs = {})
%permute_58 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_202, [0, 2, 1, 3]), kwargs = {})
%clone_52 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_58,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_89 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_52, [2, 256, 1024]), kwargs = {})
%_param_constant243 : [#users=1] = get_attr[target=_param_constant243]
%t_78 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant243,), kwargs = {})
%view_203 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_89, [512, 1024]), kwargs = {})
%_param_constant244 : [#users=1] = get_attr[target=_param_constant244]
%addmm_36 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant244, %view_203, %t_78), kwargs = {})
%view_204 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_36, [2, 256, 1024]), kwargs = {})
%add_122 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_204, %add_119), kwargs = {})
%convert_element_type_122 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_122, torch.float32), kwargs = {})
%var_mean_39 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_122, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_78 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_39, 0), kwargs = {})
%getitem_79 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_39, 1), kwargs = {})
%add_123 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_78, 1e-05), kwargs = {})
%rsqrt_39 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_123,), kwargs = {})
%sub_39 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_122, %getitem_79), kwargs = {})
%mul_86 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_39, %rsqrt_39), kwargs = {})
%_param_constant245 : [#users=1] = get_attr[target=_param_constant245]
%mul_87 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_86, %_param_constant245), kwargs = {})
%_param_constant246 : [#users=1] = get_attr[target=_param_constant246]
%add_124 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_87, %_param_constant246), kwargs = {})
%convert_element_type_123 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_124, torch.float16), kwargs = {})
%_param_constant247 : [#users=1] = get_attr[target=_param_constant247]
%t_79 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant247,), kwargs = {})
%view_205 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_123, [512, 1024]), kwargs = {})
%_param_constant248 : [#users=1] = get_attr[target=_param_constant248]
%addmm_37 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant248, %view_205, %t_79), kwargs = {})
%view_206 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%addmm_37, [2, 256, 8192]), kwargs = {})
%slice_33 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_206, -1, 0, 4096), kwargs = {})
%slice_34 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_206, -1, 4096, 8192), kwargs = {})
%gelu_5 : [#users=1] = call_function[target=torch.ops.aten.gelu](args = (%slice_34,), kwargs = {})
%mul_88 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%slice_33, %gelu_5), kwargs = {})
%_param_constant249 : [#users=1] = get_attr[target=_param_constant249]
%t_80 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant249,), kwargs = {})
%view_207 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_88, [512, 4096]), kwargs = {})
%_param_constant250 : [#users=1] = get_attr[target=_param_constant250]
%addmm_38 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant250, %view_207, %t_80), kwargs = {})
%view_208 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_38, [2, 256, 1024]), kwargs = {})
%add_125 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%view_208, %add_122), kwargs = {})
%_param_constant251 : [#users=1] = get_attr[target=_param_constant251]
%t_81 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant251,), kwargs = {})
%view_209 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%add_125, [512, 1024]), kwargs = {})
%_param_constant252 : [#users=1] = get_attr[target=_param_constant252]
%addmm_39 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant252, %view_209, %t_81), kwargs = {})
%view_210 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_39, [2, 256, 1024]), kwargs = {})
%view_211 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%view_210, [2, 16, 16, 1024]), kwargs = {})
%permute_59 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_211, [0, 3, 1, 2]), kwargs = {})
%clone_53 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_59,), kwargs = {memory_format: torch.contiguous_format})
%add_126 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%clone_53, %div_8), kwargs = {})
%view_212 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_126, [2, 32, 32, 256]), kwargs = {})
%convert_element_type_124 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_212, torch.float32), kwargs = {})
%var_mean_40 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_124, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_80 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_40, 0), kwargs = {})
%getitem_81 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_40, 1), kwargs = {})
%add_127 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_80, 1e-05), kwargs = {})
%rsqrt_40 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_127,), kwargs = {})
%sub_40 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_212, %getitem_81), kwargs = {})
%mul_89 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_40, %rsqrt_40), kwargs = {})
%view_213 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_89, [2, 1024, 16, 16]), kwargs = {})
%_param_constant253 : [#users=1] = get_attr[target=_param_constant253]
%unsqueeze_150 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant253, 0), kwargs = {})
%unsqueeze_151 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_150, 2), kwargs = {})
%unsqueeze_152 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_151, 3), kwargs = {})
%_param_constant254 : [#users=1] = get_attr[target=_param_constant254]
%unsqueeze_153 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant254, 0), kwargs = {})
%unsqueeze_154 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_153, 2), kwargs = {})
%unsqueeze_155 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_154, 3), kwargs = {})
%mul_90 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_213, %unsqueeze_155), kwargs = {})
%add_128 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_90, %unsqueeze_152), kwargs = {})
%convert_element_type_125 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_128, torch.float16), kwargs = {})
%convert_element_type_126 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_81, torch.float16), kwargs = {})
%convert_element_type_127 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_40, torch.float16), kwargs = {})
%squeeze_88 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_126, 3), kwargs = {})
%squeeze_89 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_88, 2), kwargs = {})
%squeeze_90 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_127, 3), kwargs = {})
%squeeze_91 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_90, 2), kwargs = {})
%detach_56 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_89,), kwargs = {})
%detach_57 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_91,), kwargs = {})
%silu_25 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_125,), kwargs = {})
%_param_constant255 : [#users=1] = get_attr[target=_param_constant255]
%_param_constant256 : [#users=1] = get_attr[target=_param_constant256]
%convolution_22 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_25, %_param_constant255, %_param_constant256, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_26 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant257 : [#users=1] = get_attr[target=_param_constant257]
%t_82 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant257,), kwargs = {})
%_param_constant258 : [#users=1] = get_attr[target=_param_constant258]
%addmm_40 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant258, %silu_26, %t_82), kwargs = {})
%slice_35 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_40, 0, 0, 9223372036854775807), kwargs = {})
%slice_36 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_35, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_156 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_36, 2), kwargs = {})
%unsqueeze_157 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_156, 3), kwargs = {})
%add_129 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_22, %unsqueeze_157), kwargs = {})
%view_214 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_129, [2, 32, 32, 256]), kwargs = {})
%convert_element_type_128 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_214, torch.float32), kwargs = {})
%var_mean_41 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_128, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_82 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_41, 0), kwargs = {})
%getitem_83 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_41, 1), kwargs = {})
%add_130 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_82, 1e-05), kwargs = {})
%rsqrt_41 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_130,), kwargs = {})
%sub_41 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_214, %getitem_83), kwargs = {})
%mul_91 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_41, %rsqrt_41), kwargs = {})
%view_215 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_91, [2, 1024, 16, 16]), kwargs = {})
%_param_constant259 : [#users=1] = get_attr[target=_param_constant259]
%unsqueeze_158 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant259, 0), kwargs = {})
%unsqueeze_159 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_158, 2), kwargs = {})
%unsqueeze_160 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_159, 3), kwargs = {})
%_param_constant260 : [#users=1] = get_attr[target=_param_constant260]
%unsqueeze_161 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant260, 0), kwargs = {})
%unsqueeze_162 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_161, 2), kwargs = {})
%unsqueeze_163 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_162, 3), kwargs = {})
%mul_92 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_215, %unsqueeze_163), kwargs = {})
%add_131 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_92, %unsqueeze_160), kwargs = {})
%convert_element_type_129 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_131, torch.float16), kwargs = {})
%convert_element_type_130 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_83, torch.float16), kwargs = {})
%convert_element_type_131 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_41, torch.float16), kwargs = {})
%squeeze_92 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_130, 3), kwargs = {})
%squeeze_93 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_92, 2), kwargs = {})
%squeeze_94 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_131, 3), kwargs = {})
%squeeze_95 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_94, 2), kwargs = {})
%detach_58 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_93,), kwargs = {})
%detach_59 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_95,), kwargs = {})
%silu_27 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_129,), kwargs = {})
%_param_constant261 : [#users=1] = get_attr[target=_param_constant261]
%_param_constant262 : [#users=1] = get_attr[target=_param_constant262]
%convolution_23 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_27, %_param_constant261, %_param_constant262, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%add_132 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%add_126, %convolution_23), kwargs = {})
%div_9 : [#users=2] = call_function[target=torch.ops.aten.div](args = (%add_132, 1), kwargs = {})
%view_216 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%div_9, [2, 32, 32, 256]), kwargs = {})
%convert_element_type_132 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_216, torch.float32), kwargs = {})
%var_mean_42 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_132, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_84 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_42, 0), kwargs = {})
%getitem_85 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_42, 1), kwargs = {})
%add_133 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_84, 1e-06), kwargs = {})
%rsqrt_42 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_133,), kwargs = {})
%sub_42 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_216, %getitem_85), kwargs = {})
%mul_93 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_42, %rsqrt_42), kwargs = {})
%view_217 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_93, [2, 1024, 16, 16]), kwargs = {})
%_param_constant263 : [#users=1] = get_attr[target=_param_constant263]
%unsqueeze_164 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant263, 0), kwargs = {})
%unsqueeze_165 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_164, 2), kwargs = {})
%unsqueeze_166 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_165, 3), kwargs = {})
%_param_constant264 : [#users=1] = get_attr[target=_param_constant264]
%unsqueeze_167 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant264, 0), kwargs = {})
%unsqueeze_168 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_167, 2), kwargs = {})
%unsqueeze_169 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_168, 3), kwargs = {})
%mul_94 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_217, %unsqueeze_169), kwargs = {})
%add_134 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_94, %unsqueeze_166), kwargs = {})
%convert_element_type_133 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_134, torch.float16), kwargs = {})
%convert_element_type_134 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_85, torch.float16), kwargs = {})
%convert_element_type_135 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_42, torch.float16), kwargs = {})
%squeeze_96 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_134, 3), kwargs = {})
%squeeze_97 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_96, 2), kwargs = {})
%squeeze_98 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_135, 3), kwargs = {})
%squeeze_99 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_98, 2), kwargs = {})
%detach_60 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_97,), kwargs = {})
%detach_61 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_99,), kwargs = {})
%permute_60 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%convert_element_type_133, [0, 2, 3, 1]), kwargs = {})
%view_218 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%permute_60, [2, 256, 1024]), kwargs = {})
%_param_constant265 : [#users=1] = get_attr[target=_param_constant265]
%t_83 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant265,), kwargs = {})
%expand_13 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%view_218, [2, 256, 1024]), kwargs = {})
%view_219 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_13, [2, 256, 1024]), kwargs = {})
%expand_14 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%t_83, [2, 1024, 1024]), kwargs = {})
%view_220 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_14, [2, 1024, 1024]), kwargs = {})
%bmm_18 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%view_219, %view_220), kwargs = {})
%_unsafe_view_90 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%bmm_18, [2, 256, 1024]), kwargs = {})
%_param_constant266 : [#users=1] = get_attr[target=_param_constant266]
%add_135 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%_unsafe_view_90, %_param_constant266), kwargs = {})
%convert_element_type_136 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_135, torch.float32), kwargs = {})
%var_mean_43 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_136, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_86 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_43, 0), kwargs = {})
%getitem_87 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_43, 1), kwargs = {})
%add_136 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_86, 1e-05), kwargs = {})
%rsqrt_43 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_136,), kwargs = {})
%sub_43 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_135, %getitem_87), kwargs = {})
%mul_95 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_43, %rsqrt_43), kwargs = {})
%_param_constant267 : [#users=1] = get_attr[target=_param_constant267]
%mul_96 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_95, %_param_constant267), kwargs = {})
%_param_constant268 : [#users=1] = get_attr[target=_param_constant268]
%add_137 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_96, %_param_constant268), kwargs = {})
%convert_element_type_137 : [#users=3] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_137, torch.float16), kwargs = {})
%_param_constant269 : [#users=1] = get_attr[target=_param_constant269]
%t_84 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant269,), kwargs = {})
%view_221 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_137, [512, 1024]), kwargs = {})
%mm_36 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_221, %t_84), kwargs = {})
%_unsafe_view_91 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_36, [2, 256, 1024]), kwargs = {})
%view_222 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_91, [2, 256, 8, 128]), kwargs = {})
%permute_61 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_222, [0, 2, 1, 3]), kwargs = {})
%clone_54 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_61,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_92 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_54, [16, 256, 128]), kwargs = {})
%_param_constant270 : [#users=1] = get_attr[target=_param_constant270]
%t_85 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant270,), kwargs = {})
%view_223 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_137, [512, 1024]), kwargs = {})
%mm_37 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_223, %t_85), kwargs = {})
%_unsafe_view_93 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_37, [2, 256, 1024]), kwargs = {})
%_param_constant271 : [#users=1] = get_attr[target=_param_constant271]
%t_86 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant271,), kwargs = {})
%view_224 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_137, [512, 1024]), kwargs = {})
%mm_38 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_224, %t_86), kwargs = {})
%_unsafe_view_94 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_38, [2, 256, 1024]), kwargs = {})
%view_225 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_93, [2, 256, 8, 128]), kwargs = {})
%permute_62 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_225, [0, 2, 1, 3]), kwargs = {})
%clone_55 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_62,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_95 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_55, [16, 256, 128]), kwargs = {})
%view_226 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_94, [2, 256, 8, 128]), kwargs = {})
%permute_63 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_226, [0, 2, 1, 3]), kwargs = {})
%clone_56 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_63,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_96 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_56, [16, 256, 128]), kwargs = {})
%empty_12 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 256, 256],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_12 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_95, -1, -2), kwargs = {})
%baddbmm_12 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_12, %_unsafe_view_92, %transpose_12), kwargs = {beta: 0, alpha: 0.08838834764831845})
%_softmax_12 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_12, -1, False), kwargs = {})
%detach_62 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_12,), kwargs = {})
%bmm_19 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_12, %_unsafe_view_96), kwargs = {})
%view_227 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_19, [2, 8, 256, 128]), kwargs = {})
%permute_64 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_227, [0, 2, 1, 3]), kwargs = {})
%clone_57 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_64,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_97 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_57, [2, 256, 1024]), kwargs = {})
%_param_constant272 : [#users=1] = get_attr[target=_param_constant272]
%t_87 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant272,), kwargs = {})
%view_228 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_97, [512, 1024]), kwargs = {})
%_param_constant273 : [#users=1] = get_attr[target=_param_constant273]
%addmm_41 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant273, %view_228, %t_87), kwargs = {})
%view_229 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_41, [2, 256, 1024]), kwargs = {})
%add_138 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_229, %add_135), kwargs = {})
%convert_element_type_138 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_138, torch.float32), kwargs = {})
%var_mean_44 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_138, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_88 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_44, 0), kwargs = {})
%getitem_89 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_44, 1), kwargs = {})
%add_139 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_88, 1e-05), kwargs = {})
%rsqrt_44 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_139,), kwargs = {})
%sub_44 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_138, %getitem_89), kwargs = {})
%mul_97 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_44, %rsqrt_44), kwargs = {})
%_param_constant274 : [#users=1] = get_attr[target=_param_constant274]
%mul_98 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_97, %_param_constant274), kwargs = {})
%_param_constant275 : [#users=1] = get_attr[target=_param_constant275]
%add_140 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_98, %_param_constant275), kwargs = {})
%convert_element_type_139 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_140, torch.float16), kwargs = {})
%_param_constant276 : [#users=1] = get_attr[target=_param_constant276]
%t_88 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant276,), kwargs = {})
%view_230 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_139, [512, 1024]), kwargs = {})
%mm_39 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_230, %t_88), kwargs = {})
%_unsafe_view_98 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_39, [2, 256, 1024]), kwargs = {})
%view_231 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_98, [2, 256, 8, 128]), kwargs = {})
%permute_65 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_231, [0, 2, 1, 3]), kwargs = {})
%clone_58 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_65,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_99 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_58, [16, 256, 128]), kwargs = {})
%_param_constant277 : [#users=1] = get_attr[target=_param_constant277]
%t_89 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant277,), kwargs = {})
%view_232 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_40 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_232, %t_89), kwargs = {})
%_unsafe_view_100 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_40, [2, 77, 1024]), kwargs = {})
%_param_constant278 : [#users=1] = get_attr[target=_param_constant278]
%t_90 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant278,), kwargs = {})
%view_233 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_41 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_233, %t_90), kwargs = {})
%_unsafe_view_101 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_41, [2, 77, 1024]), kwargs = {})
%view_234 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_100, [2, 77, 8, 128]), kwargs = {})
%permute_66 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_234, [0, 2, 1, 3]), kwargs = {})
%clone_59 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_66,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_102 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_59, [16, 77, 128]), kwargs = {})
%view_235 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_101, [2, 77, 8, 128]), kwargs = {})
%permute_67 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_235, [0, 2, 1, 3]), kwargs = {})
%clone_60 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_67,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_103 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_60, [16, 77, 128]), kwargs = {})
%empty_13 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 256, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_13 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_102, -1, -2), kwargs = {})
%baddbmm_13 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_13, %_unsafe_view_99, %transpose_13), kwargs = {beta: 0, alpha: 0.08838834764831845})
%_softmax_13 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_13, -1, False), kwargs = {})
%detach_63 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_13,), kwargs = {})
%bmm_20 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_13, %_unsafe_view_103), kwargs = {})
%view_236 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_20, [2, 8, 256, 128]), kwargs = {})
%permute_68 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_236, [0, 2, 1, 3]), kwargs = {})
%clone_61 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_68,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_104 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_61, [2, 256, 1024]), kwargs = {})
%_param_constant279 : [#users=1] = get_attr[target=_param_constant279]
%t_91 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant279,), kwargs = {})
%view_237 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_104, [512, 1024]), kwargs = {})
%_param_constant280 : [#users=1] = get_attr[target=_param_constant280]
%addmm_42 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant280, %view_237, %t_91), kwargs = {})
%view_238 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_42, [2, 256, 1024]), kwargs = {})
%add_141 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_238, %add_138), kwargs = {})
%convert_element_type_140 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_141, torch.float32), kwargs = {})
%var_mean_45 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_140, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_90 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_45, 0), kwargs = {})
%getitem_91 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_45, 1), kwargs = {})
%add_142 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_90, 1e-05), kwargs = {})
%rsqrt_45 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_142,), kwargs = {})
%sub_45 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_141, %getitem_91), kwargs = {})
%mul_99 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_45, %rsqrt_45), kwargs = {})
%_param_constant281 : [#users=1] = get_attr[target=_param_constant281]
%mul_100 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_99, %_param_constant281), kwargs = {})
%_param_constant282 : [#users=1] = get_attr[target=_param_constant282]
%add_143 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_100, %_param_constant282), kwargs = {})
%convert_element_type_141 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_143, torch.float16), kwargs = {})
%_param_constant283 : [#users=1] = get_attr[target=_param_constant283]
%t_92 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant283,), kwargs = {})
%view_239 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_141, [512, 1024]), kwargs = {})
%_param_constant284 : [#users=1] = get_attr[target=_param_constant284]
%addmm_43 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant284, %view_239, %t_92), kwargs = {})
%view_240 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%addmm_43, [2, 256, 8192]), kwargs = {})
%slice_37 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_240, -1, 0, 4096), kwargs = {})
%slice_38 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_240, -1, 4096, 8192), kwargs = {})
%gelu_6 : [#users=1] = call_function[target=torch.ops.aten.gelu](args = (%slice_38,), kwargs = {})
%mul_101 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%slice_37, %gelu_6), kwargs = {})
%_param_constant285 : [#users=1] = get_attr[target=_param_constant285]
%t_93 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant285,), kwargs = {})
%view_241 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_101, [512, 4096]), kwargs = {})
%_param_constant286 : [#users=1] = get_attr[target=_param_constant286]
%addmm_44 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant286, %view_241, %t_93), kwargs = {})
%view_242 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_44, [2, 256, 1024]), kwargs = {})
%add_144 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%view_242, %add_141), kwargs = {})
%_param_constant287 : [#users=1] = get_attr[target=_param_constant287]
%t_94 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant287,), kwargs = {})
%view_243 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%add_144, [512, 1024]), kwargs = {})
%_param_constant288 : [#users=1] = get_attr[target=_param_constant288]
%addmm_45 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant288, %view_243, %t_94), kwargs = {})
%view_244 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_45, [2, 256, 1024]), kwargs = {})
%view_245 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%view_244, [2, 16, 16, 1024]), kwargs = {})
%permute_69 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_245, [0, 3, 1, 2]), kwargs = {})
%clone_62 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_69,), kwargs = {memory_format: torch.contiguous_format})
%add_145 : [#users=2] = call_function[target=torch.ops.aten.add](args = (%clone_62, %div_9), kwargs = {})
%view_246 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_145, [2, 32, 32, 256]), kwargs = {})
%convert_element_type_142 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_246, torch.float32), kwargs = {})
%var_mean_46 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_142, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_92 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_46, 0), kwargs = {})
%getitem_93 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_46, 1), kwargs = {})
%add_146 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_92, 1e-05), kwargs = {})
%rsqrt_46 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_146,), kwargs = {})
%sub_46 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_246, %getitem_93), kwargs = {})
%mul_102 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_46, %rsqrt_46), kwargs = {})
%view_247 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_102, [2, 1024, 16, 16]), kwargs = {})
%_param_constant289 : [#users=1] = get_attr[target=_param_constant289]
%unsqueeze_170 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant289, 0), kwargs = {})
%unsqueeze_171 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_170, 2), kwargs = {})
%unsqueeze_172 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_171, 3), kwargs = {})
%_param_constant290 : [#users=1] = get_attr[target=_param_constant290]
%unsqueeze_173 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant290, 0), kwargs = {})
%unsqueeze_174 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_173, 2), kwargs = {})
%unsqueeze_175 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_174, 3), kwargs = {})
%mul_103 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_247, %unsqueeze_175), kwargs = {})
%add_147 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_103, %unsqueeze_172), kwargs = {})
%convert_element_type_143 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_147, torch.float16), kwargs = {})
%convert_element_type_144 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_93, torch.float16), kwargs = {})
%convert_element_type_145 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_46, torch.float16), kwargs = {})
%squeeze_100 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_144, 3), kwargs = {})
%squeeze_101 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_100, 2), kwargs = {})
%squeeze_102 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_145, 3), kwargs = {})
%squeeze_103 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_102, 2), kwargs = {})
%detach_64 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_101,), kwargs = {})
%detach_65 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_103,), kwargs = {})
%silu_28 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_143,), kwargs = {})
%_param_constant291 : [#users=1] = get_attr[target=_param_constant291]
%_param_constant292 : [#users=1] = get_attr[target=_param_constant292]
%convolution_24 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_28, %_param_constant291, %_param_constant292, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_29 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant293 : [#users=1] = get_attr[target=_param_constant293]
%t_95 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant293,), kwargs = {})
%_param_constant294 : [#users=1] = get_attr[target=_param_constant294]
%addmm_46 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant294, %silu_29, %t_95), kwargs = {})
%slice_39 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_46, 0, 0, 9223372036854775807), kwargs = {})
%slice_40 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_39, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_176 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_40, 2), kwargs = {})
%unsqueeze_177 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_176, 3), kwargs = {})
%add_148 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_24, %unsqueeze_177), kwargs = {})
%view_248 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_148, [2, 32, 32, 256]), kwargs = {})
%convert_element_type_146 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_248, torch.float32), kwargs = {})
%var_mean_47 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_146, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_94 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_47, 0), kwargs = {})
%getitem_95 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_47, 1), kwargs = {})
%add_149 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_94, 1e-05), kwargs = {})
%rsqrt_47 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_149,), kwargs = {})
%sub_47 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_248, %getitem_95), kwargs = {})
%mul_104 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_47, %rsqrt_47), kwargs = {})
%view_249 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_104, [2, 1024, 16, 16]), kwargs = {})
%_param_constant295 : [#users=1] = get_attr[target=_param_constant295]
%unsqueeze_178 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant295, 0), kwargs = {})
%unsqueeze_179 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_178, 2), kwargs = {})
%unsqueeze_180 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_179, 3), kwargs = {})
%_param_constant296 : [#users=1] = get_attr[target=_param_constant296]
%unsqueeze_181 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant296, 0), kwargs = {})
%unsqueeze_182 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_181, 2), kwargs = {})
%unsqueeze_183 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_182, 3), kwargs = {})
%mul_105 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_249, %unsqueeze_183), kwargs = {})
%add_150 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_105, %unsqueeze_180), kwargs = {})
%convert_element_type_147 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_150, torch.float16), kwargs = {})
%convert_element_type_148 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_95, torch.float16), kwargs = {})
%convert_element_type_149 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_47, torch.float16), kwargs = {})
%squeeze_104 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_148, 3), kwargs = {})
%squeeze_105 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_104, 2), kwargs = {})
%squeeze_106 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_149, 3), kwargs = {})
%squeeze_107 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_106, 2), kwargs = {})
%detach_66 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_105,), kwargs = {})
%detach_67 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_107,), kwargs = {})
%silu_30 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_147,), kwargs = {})
%_param_constant297 : [#users=1] = get_attr[target=_param_constant297]
%_param_constant298 : [#users=1] = get_attr[target=_param_constant298]
%convolution_25 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_30, %_param_constant297, %_param_constant298, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%add_151 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%add_145, %convolution_25), kwargs = {})
%div_10 : [#users=1] = call_function[target=torch.ops.aten.div](args = (%add_151, 1), kwargs = {})
%cat_2 : [#users=2] = call_function[target=torch.ops.aten.cat](args = ([%div_10, %add_126], 1), kwargs = {})
%view_250 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%cat_2, [2, 32, 64, 256]), kwargs = {})
%convert_element_type_150 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_250, torch.float32), kwargs = {})
%var_mean_48 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_150, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_96 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_48, 0), kwargs = {})
%getitem_97 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_48, 1), kwargs = {})
%add_152 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_96, 1e-05), kwargs = {})
%rsqrt_48 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_152,), kwargs = {})
%sub_48 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_250, %getitem_97), kwargs = {})
%mul_106 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_48, %rsqrt_48), kwargs = {})
%view_251 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_106, [2, 2048, 16, 16]), kwargs = {})
%_param_constant299 : [#users=1] = get_attr[target=_param_constant299]
%unsqueeze_184 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant299, 0), kwargs = {})
%unsqueeze_185 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_184, 2), kwargs = {})
%unsqueeze_186 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_185, 3), kwargs = {})
%_param_constant300 : [#users=1] = get_attr[target=_param_constant300]
%unsqueeze_187 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant300, 0), kwargs = {})
%unsqueeze_188 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_187, 2), kwargs = {})
%unsqueeze_189 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_188, 3), kwargs = {})
%mul_107 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_251, %unsqueeze_189), kwargs = {})
%add_153 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_107, %unsqueeze_186), kwargs = {})
%convert_element_type_151 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_153, torch.float16), kwargs = {})
%convert_element_type_152 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_97, torch.float16), kwargs = {})
%convert_element_type_153 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_48, torch.float16), kwargs = {})
%squeeze_108 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_152, 3), kwargs = {})
%squeeze_109 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_108, 2), kwargs = {})
%squeeze_110 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_153, 3), kwargs = {})
%squeeze_111 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_110, 2), kwargs = {})
%detach_68 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_109,), kwargs = {})
%detach_69 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_111,), kwargs = {})
%silu_31 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_151,), kwargs = {})
%_param_constant301 : [#users=1] = get_attr[target=_param_constant301]
%_param_constant302 : [#users=1] = get_attr[target=_param_constant302]
%convolution_26 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_31, %_param_constant301, %_param_constant302, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_32 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant303 : [#users=1] = get_attr[target=_param_constant303]
%t_96 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant303,), kwargs = {})
%_param_constant304 : [#users=1] = get_attr[target=_param_constant304]
%addmm_47 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant304, %silu_32, %t_96), kwargs = {})
%slice_41 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_47, 0, 0, 9223372036854775807), kwargs = {})
%slice_42 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_41, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_190 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_42, 2), kwargs = {})
%unsqueeze_191 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_190, 3), kwargs = {})
%add_154 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_26, %unsqueeze_191), kwargs = {})
%view_252 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_154, [2, 32, 32, 256]), kwargs = {})
%convert_element_type_154 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_252, torch.float32), kwargs = {})
%var_mean_49 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_154, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_98 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_49, 0), kwargs = {})
%getitem_99 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_49, 1), kwargs = {})
%add_155 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_98, 1e-05), kwargs = {})
%rsqrt_49 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_155,), kwargs = {})
%sub_49 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_252, %getitem_99), kwargs = {})
%mul_108 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_49, %rsqrt_49), kwargs = {})
%view_253 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_108, [2, 1024, 16, 16]), kwargs = {})
%_param_constant305 : [#users=1] = get_attr[target=_param_constant305]
%unsqueeze_192 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant305, 0), kwargs = {})
%unsqueeze_193 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_192, 2), kwargs = {})
%unsqueeze_194 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_193, 3), kwargs = {})
%_param_constant306 : [#users=1] = get_attr[target=_param_constant306]
%unsqueeze_195 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant306, 0), kwargs = {})
%unsqueeze_196 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_195, 2), kwargs = {})
%unsqueeze_197 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_196, 3), kwargs = {})
%mul_109 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_253, %unsqueeze_197), kwargs = {})
%add_156 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_109, %unsqueeze_194), kwargs = {})
%convert_element_type_155 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_156, torch.float16), kwargs = {})
%convert_element_type_156 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_99, torch.float16), kwargs = {})
%convert_element_type_157 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_49, torch.float16), kwargs = {})
%squeeze_112 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_156, 3), kwargs = {})
%squeeze_113 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_112, 2), kwargs = {})
%squeeze_114 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_157, 3), kwargs = {})
%squeeze_115 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_114, 2), kwargs = {})
%detach_70 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_113,), kwargs = {})
%detach_71 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_115,), kwargs = {})
%silu_33 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_155,), kwargs = {})
%_param_constant307 : [#users=1] = get_attr[target=_param_constant307]
%_param_constant308 : [#users=1] = get_attr[target=_param_constant308]
%convolution_27 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_33, %_param_constant307, %_param_constant308, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%_param_constant309 : [#users=1] = get_attr[target=_param_constant309]
%_param_constant310 : [#users=1] = get_attr[target=_param_constant310]
%convolution_28 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%cat_2, %_param_constant309, %_param_constant310, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
%add_157 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_28, %convolution_27), kwargs = {})
%div_11 : [#users=2] = call_function[target=torch.ops.aten.div](args = (%add_157, 1.0), kwargs = {})
%view_254 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%div_11, [2, 32, 32, 256]), kwargs = {})
%convert_element_type_158 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_254, torch.float32), kwargs = {})
%var_mean_50 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_158, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_100 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_50, 0), kwargs = {})
%getitem_101 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_50, 1), kwargs = {})
%add_158 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_100, 1e-06), kwargs = {})
%rsqrt_50 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_158,), kwargs = {})
%sub_50 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_254, %getitem_101), kwargs = {})
%mul_110 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_50, %rsqrt_50), kwargs = {})
%view_255 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_110, [2, 1024, 16, 16]), kwargs = {})
%_param_constant311 : [#users=1] = get_attr[target=_param_constant311]
%unsqueeze_198 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant311, 0), kwargs = {})
%unsqueeze_199 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_198, 2), kwargs = {})
%unsqueeze_200 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_199, 3), kwargs = {})
%_param_constant312 : [#users=1] = get_attr[target=_param_constant312]
%unsqueeze_201 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant312, 0), kwargs = {})
%unsqueeze_202 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_201, 2), kwargs = {})
%unsqueeze_203 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_202, 3), kwargs = {})
%mul_111 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_255, %unsqueeze_203), kwargs = {})
%add_159 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_111, %unsqueeze_200), kwargs = {})
%convert_element_type_159 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_159, torch.float16), kwargs = {})
%convert_element_type_160 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_101, torch.float16), kwargs = {})
%convert_element_type_161 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_50, torch.float16), kwargs = {})
%squeeze_116 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_160, 3), kwargs = {})
%squeeze_117 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_116, 2), kwargs = {})
%squeeze_118 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_161, 3), kwargs = {})
%squeeze_119 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_118, 2), kwargs = {})
%detach_72 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_117,), kwargs = {})
%detach_73 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_119,), kwargs = {})
%permute_70 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%convert_element_type_159, [0, 2, 3, 1]), kwargs = {})
%view_256 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%permute_70, [2, 256, 1024]), kwargs = {})
%_param_constant313 : [#users=1] = get_attr[target=_param_constant313]
%t_97 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant313,), kwargs = {})
%expand_15 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%view_256, [2, 256, 1024]), kwargs = {})
%view_257 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_15, [2, 256, 1024]), kwargs = {})
%expand_16 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%t_97, [2, 1024, 1024]), kwargs = {})
%view_258 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_16, [2, 1024, 1024]), kwargs = {})
%bmm_21 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%view_257, %view_258), kwargs = {})
%_unsafe_view_105 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%bmm_21, [2, 256, 1024]), kwargs = {})
%_param_constant314 : [#users=1] = get_attr[target=_param_constant314]
%add_160 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%_unsafe_view_105, %_param_constant314), kwargs = {})
%convert_element_type_162 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_160, torch.float32), kwargs = {})
%var_mean_51 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_162, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_102 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_51, 0), kwargs = {})
%getitem_103 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_51, 1), kwargs = {})
%add_161 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_102, 1e-05), kwargs = {})
%rsqrt_51 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_161,), kwargs = {})
%sub_51 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_160, %getitem_103), kwargs = {})
%mul_112 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_51, %rsqrt_51), kwargs = {})
%_param_constant315 : [#users=1] = get_attr[target=_param_constant315]
%mul_113 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_112, %_param_constant315), kwargs = {})
%_param_constant316 : [#users=1] = get_attr[target=_param_constant316]
%add_162 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_113, %_param_constant316), kwargs = {})
%convert_element_type_163 : [#users=3] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_162, torch.float16), kwargs = {})
%_param_constant317 : [#users=1] = get_attr[target=_param_constant317]
%t_98 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant317,), kwargs = {})
%view_259 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_163, [512, 1024]), kwargs = {})
%mm_42 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_259, %t_98), kwargs = {})
%_unsafe_view_106 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_42, [2, 256, 1024]), kwargs = {})
%view_260 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_106, [2, 256, 8, 128]), kwargs = {})
%permute_71 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_260, [0, 2, 1, 3]), kwargs = {})
%clone_63 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_71,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_107 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_63, [16, 256, 128]), kwargs = {})
%_param_constant318 : [#users=1] = get_attr[target=_param_constant318]
%t_99 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant318,), kwargs = {})
%view_261 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_163, [512, 1024]), kwargs = {})
%mm_43 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_261, %t_99), kwargs = {})
%_unsafe_view_108 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_43, [2, 256, 1024]), kwargs = {})
%_param_constant319 : [#users=1] = get_attr[target=_param_constant319]
%t_100 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant319,), kwargs = {})
%view_262 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_163, [512, 1024]), kwargs = {})
%mm_44 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_262, %t_100), kwargs = {})
%_unsafe_view_109 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_44, [2, 256, 1024]), kwargs = {})
%view_263 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_108, [2, 256, 8, 128]), kwargs = {})
%permute_72 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_263, [0, 2, 1, 3]), kwargs = {})
%clone_64 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_72,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_110 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_64, [16, 256, 128]), kwargs = {})
%view_264 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_109, [2, 256, 8, 128]), kwargs = {})
%permute_73 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_264, [0, 2, 1, 3]), kwargs = {})
%clone_65 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_73,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_111 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_65, [16, 256, 128]), kwargs = {})
%empty_14 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 256, 256],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_14 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_110, -1, -2), kwargs = {})
%baddbmm_14 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_14, %_unsafe_view_107, %transpose_14), kwargs = {beta: 0, alpha: 0.08838834764831845})
%_softmax_14 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_14, -1, False), kwargs = {})
%detach_74 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_14,), kwargs = {})
%bmm_22 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_14, %_unsafe_view_111), kwargs = {})
%view_265 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_22, [2, 8, 256, 128]), kwargs = {})
%permute_74 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_265, [0, 2, 1, 3]), kwargs = {})
%clone_66 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_74,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_112 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_66, [2, 256, 1024]), kwargs = {})
%_param_constant320 : [#users=1] = get_attr[target=_param_constant320]
%t_101 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant320,), kwargs = {})
%view_266 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_112, [512, 1024]), kwargs = {})
%_param_constant321 : [#users=1] = get_attr[target=_param_constant321]
%addmm_48 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant321, %view_266, %t_101), kwargs = {})
%view_267 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_48, [2, 256, 1024]), kwargs = {})
%add_163 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_267, %add_160), kwargs = {})
%convert_element_type_164 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_163, torch.float32), kwargs = {})
%var_mean_52 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_164, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_104 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_52, 0), kwargs = {})
%getitem_105 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_52, 1), kwargs = {})
%add_164 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_104, 1e-05), kwargs = {})
%rsqrt_52 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_164,), kwargs = {})
%sub_52 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_163, %getitem_105), kwargs = {})
%mul_114 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_52, %rsqrt_52), kwargs = {})
%_param_constant322 : [#users=1] = get_attr[target=_param_constant322]
%mul_115 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_114, %_param_constant322), kwargs = {})
%_param_constant323 : [#users=1] = get_attr[target=_param_constant323]
%add_165 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_115, %_param_constant323), kwargs = {})
%convert_element_type_165 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_165, torch.float16), kwargs = {})
%_param_constant324 : [#users=1] = get_attr[target=_param_constant324]
%t_102 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant324,), kwargs = {})
%view_268 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_165, [512, 1024]), kwargs = {})
%mm_45 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_268, %t_102), kwargs = {})
%_unsafe_view_113 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_45, [2, 256, 1024]), kwargs = {})
%view_269 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_113, [2, 256, 8, 128]), kwargs = {})
%permute_75 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_269, [0, 2, 1, 3]), kwargs = {})
%clone_67 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_75,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_114 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_67, [16, 256, 128]), kwargs = {})
%_param_constant325 : [#users=1] = get_attr[target=_param_constant325]
%t_103 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant325,), kwargs = {})
%view_270 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_46 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_270, %t_103), kwargs = {})
%_unsafe_view_115 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_46, [2, 77, 1024]), kwargs = {})
%_param_constant326 : [#users=1] = get_attr[target=_param_constant326]
%t_104 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant326,), kwargs = {})
%view_271 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_47 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_271, %t_104), kwargs = {})
%_unsafe_view_116 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_47, [2, 77, 1024]), kwargs = {})
%view_272 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_115, [2, 77, 8, 128]), kwargs = {})
%permute_76 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_272, [0, 2, 1, 3]), kwargs = {})
%clone_68 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_76,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_117 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_68, [16, 77, 128]), kwargs = {})
%view_273 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_116, [2, 77, 8, 128]), kwargs = {})
%permute_77 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_273, [0, 2, 1, 3]), kwargs = {})
%clone_69 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_77,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_118 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_69, [16, 77, 128]), kwargs = {})
%empty_15 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 256, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_15 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_117, -1, -2), kwargs = {})
%baddbmm_15 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_15, %_unsafe_view_114, %transpose_15), kwargs = {beta: 0, alpha: 0.08838834764831845})
%_softmax_15 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_15, -1, False), kwargs = {})
%detach_75 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_15,), kwargs = {})
%bmm_23 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_15, %_unsafe_view_118), kwargs = {})
%view_274 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_23, [2, 8, 256, 128]), kwargs = {})
%permute_78 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_274, [0, 2, 1, 3]), kwargs = {})
%clone_70 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_78,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_119 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_70, [2, 256, 1024]), kwargs = {})
%_param_constant327 : [#users=1] = get_attr[target=_param_constant327]
%t_105 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant327,), kwargs = {})
%view_275 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_119, [512, 1024]), kwargs = {})
%_param_constant328 : [#users=1] = get_attr[target=_param_constant328]
%addmm_49 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant328, %view_275, %t_105), kwargs = {})
%view_276 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_49, [2, 256, 1024]), kwargs = {})
%add_166 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_276, %add_163), kwargs = {})
%convert_element_type_166 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_166, torch.float32), kwargs = {})
%var_mean_53 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_166, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_106 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_53, 0), kwargs = {})
%getitem_107 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_53, 1), kwargs = {})
%add_167 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_106, 1e-05), kwargs = {})
%rsqrt_53 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_167,), kwargs = {})
%sub_53 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_166, %getitem_107), kwargs = {})
%mul_116 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_53, %rsqrt_53), kwargs = {})
%_param_constant329 : [#users=1] = get_attr[target=_param_constant329]
%mul_117 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_116, %_param_constant329), kwargs = {})
%_param_constant330 : [#users=1] = get_attr[target=_param_constant330]
%add_168 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_117, %_param_constant330), kwargs = {})
%convert_element_type_167 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_168, torch.float16), kwargs = {})
%_param_constant331 : [#users=1] = get_attr[target=_param_constant331]
%t_106 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant331,), kwargs = {})
%view_277 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_167, [512, 1024]), kwargs = {})
%_param_constant332 : [#users=1] = get_attr[target=_param_constant332]
%addmm_50 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant332, %view_277, %t_106), kwargs = {})
%view_278 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%addmm_50, [2, 256, 8192]), kwargs = {})
%slice_43 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_278, -1, 0, 4096), kwargs = {})
%slice_44 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_278, -1, 4096, 8192), kwargs = {})
%gelu_7 : [#users=1] = call_function[target=torch.ops.aten.gelu](args = (%slice_44,), kwargs = {})
%mul_118 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%slice_43, %gelu_7), kwargs = {})
%_param_constant333 : [#users=1] = get_attr[target=_param_constant333]
%t_107 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant333,), kwargs = {})
%view_279 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_118, [512, 4096]), kwargs = {})
%_param_constant334 : [#users=1] = get_attr[target=_param_constant334]
%addmm_51 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant334, %view_279, %t_107), kwargs = {})
%view_280 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_51, [2, 256, 1024]), kwargs = {})
%add_169 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%view_280, %add_166), kwargs = {})
%_param_constant335 : [#users=1] = get_attr[target=_param_constant335]
%t_108 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant335,), kwargs = {})
%view_281 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%add_169, [512, 1024]), kwargs = {})
%_param_constant336 : [#users=1] = get_attr[target=_param_constant336]
%addmm_52 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant336, %view_281, %t_108), kwargs = {})
%view_282 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_52, [2, 256, 1024]), kwargs = {})
%view_283 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%view_282, [2, 16, 16, 1024]), kwargs = {})
%permute_79 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_283, [0, 3, 1, 2]), kwargs = {})
%clone_71 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_79,), kwargs = {memory_format: torch.contiguous_format})
%add_170 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%clone_71, %div_11), kwargs = {})
%cat_3 : [#users=2] = call_function[target=torch.ops.aten.cat](args = ([%add_170, %add_107], 1), kwargs = {})
%view_284 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%cat_3, [2, 32, 64, 256]), kwargs = {})
%convert_element_type_168 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_284, torch.float32), kwargs = {})
%var_mean_54 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_168, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_108 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_54, 0), kwargs = {})
%getitem_109 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_54, 1), kwargs = {})
%add_171 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_108, 1e-05), kwargs = {})
%rsqrt_54 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_171,), kwargs = {})
%sub_54 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_284, %getitem_109), kwargs = {})
%mul_119 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_54, %rsqrt_54), kwargs = {})
%view_285 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_119, [2, 2048, 16, 16]), kwargs = {})
%_param_constant337 : [#users=1] = get_attr[target=_param_constant337]
%unsqueeze_204 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant337, 0), kwargs = {})
%unsqueeze_205 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_204, 2), kwargs = {})
%unsqueeze_206 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_205, 3), kwargs = {})
%_param_constant338 : [#users=1] = get_attr[target=_param_constant338]
%unsqueeze_207 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant338, 0), kwargs = {})
%unsqueeze_208 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_207, 2), kwargs = {})
%unsqueeze_209 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_208, 3), kwargs = {})
%mul_120 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_285, %unsqueeze_209), kwargs = {})
%add_172 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_120, %unsqueeze_206), kwargs = {})
%convert_element_type_169 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_172, torch.float16), kwargs = {})
%convert_element_type_170 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_109, torch.float16), kwargs = {})
%convert_element_type_171 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_54, torch.float16), kwargs = {})
%squeeze_120 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_170, 3), kwargs = {})
%squeeze_121 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_120, 2), kwargs = {})
%squeeze_122 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_171, 3), kwargs = {})
%squeeze_123 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_122, 2), kwargs = {})
%detach_76 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_121,), kwargs = {})
%detach_77 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_123,), kwargs = {})
%silu_34 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_169,), kwargs = {})
%_param_constant339 : [#users=1] = get_attr[target=_param_constant339]
%_param_constant340 : [#users=1] = get_attr[target=_param_constant340]
%convolution_29 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_34, %_param_constant339, %_param_constant340, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_35 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant341 : [#users=1] = get_attr[target=_param_constant341]
%t_109 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant341,), kwargs = {})
%_param_constant342 : [#users=1] = get_attr[target=_param_constant342]
%addmm_53 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant342, %silu_35, %t_109), kwargs = {})
%slice_45 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_53, 0, 0, 9223372036854775807), kwargs = {})
%slice_46 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_45, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_210 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_46, 2), kwargs = {})
%unsqueeze_211 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_210, 3), kwargs = {})
%add_173 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_29, %unsqueeze_211), kwargs = {})
%view_286 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_173, [2, 32, 32, 256]), kwargs = {})
%convert_element_type_172 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_286, torch.float32), kwargs = {})
%var_mean_55 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_172, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_110 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_55, 0), kwargs = {})
%getitem_111 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_55, 1), kwargs = {})
%add_174 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_110, 1e-05), kwargs = {})
%rsqrt_55 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_174,), kwargs = {})
%sub_55 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_286, %getitem_111), kwargs = {})
%mul_121 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_55, %rsqrt_55), kwargs = {})
%view_287 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_121, [2, 1024, 16, 16]), kwargs = {})
%_param_constant343 : [#users=1] = get_attr[target=_param_constant343]
%unsqueeze_212 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant343, 0), kwargs = {})
%unsqueeze_213 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_212, 2), kwargs = {})
%unsqueeze_214 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_213, 3), kwargs = {})
%_param_constant344 : [#users=1] = get_attr[target=_param_constant344]
%unsqueeze_215 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant344, 0), kwargs = {})
%unsqueeze_216 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_215, 2), kwargs = {})
%unsqueeze_217 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_216, 3), kwargs = {})
%mul_122 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_287, %unsqueeze_217), kwargs = {})
%add_175 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_122, %unsqueeze_214), kwargs = {})
%convert_element_type_173 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_175, torch.float16), kwargs = {})
%convert_element_type_174 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_111, torch.float16), kwargs = {})
%convert_element_type_175 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_55, torch.float16), kwargs = {})
%squeeze_124 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_174, 3), kwargs = {})
%squeeze_125 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_124, 2), kwargs = {})
%squeeze_126 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_175, 3), kwargs = {})
%squeeze_127 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_126, 2), kwargs = {})
%detach_78 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_125,), kwargs = {})
%detach_79 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_127,), kwargs = {})
%silu_36 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_173,), kwargs = {})
%_param_constant345 : [#users=1] = get_attr[target=_param_constant345]
%_param_constant346 : [#users=1] = get_attr[target=_param_constant346]
%convolution_30 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_36, %_param_constant345, %_param_constant346, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%_param_constant347 : [#users=1] = get_attr[target=_param_constant347]
%_param_constant348 : [#users=1] = get_attr[target=_param_constant348]
%convolution_31 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%cat_3, %_param_constant347, %_param_constant348, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
%add_176 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_31, %convolution_30), kwargs = {})
%div_12 : [#users=2] = call_function[target=torch.ops.aten.div](args = (%add_176, 1.0), kwargs = {})
%view_288 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%div_12, [2, 32, 32, 256]), kwargs = {})
%convert_element_type_176 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_288, torch.float32), kwargs = {})
%var_mean_56 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_176, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_112 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_56, 0), kwargs = {})
%getitem_113 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_56, 1), kwargs = {})
%add_177 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_112, 1e-06), kwargs = {})
%rsqrt_56 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_177,), kwargs = {})
%sub_56 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_288, %getitem_113), kwargs = {})
%mul_123 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_56, %rsqrt_56), kwargs = {})
%view_289 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_123, [2, 1024, 16, 16]), kwargs = {})
%_param_constant349 : [#users=1] = get_attr[target=_param_constant349]
%unsqueeze_218 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant349, 0), kwargs = {})
%unsqueeze_219 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_218, 2), kwargs = {})
%unsqueeze_220 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_219, 3), kwargs = {})
%_param_constant350 : [#users=1] = get_attr[target=_param_constant350]
%unsqueeze_221 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant350, 0), kwargs = {})
%unsqueeze_222 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_221, 2), kwargs = {})
%unsqueeze_223 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_222, 3), kwargs = {})
%mul_124 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_289, %unsqueeze_223), kwargs = {})
%add_178 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_124, %unsqueeze_220), kwargs = {})
%convert_element_type_177 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_178, torch.float16), kwargs = {})
%convert_element_type_178 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_113, torch.float16), kwargs = {})
%convert_element_type_179 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_56, torch.float16), kwargs = {})
%squeeze_128 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_178, 3), kwargs = {})
%squeeze_129 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_128, 2), kwargs = {})
%squeeze_130 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_179, 3), kwargs = {})
%squeeze_131 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_130, 2), kwargs = {})
%detach_80 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_129,), kwargs = {})
%detach_81 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_131,), kwargs = {})
%permute_80 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%convert_element_type_177, [0, 2, 3, 1]), kwargs = {})
%view_290 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%permute_80, [2, 256, 1024]), kwargs = {})
%_param_constant351 : [#users=1] = get_attr[target=_param_constant351]
%t_110 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant351,), kwargs = {})
%expand_17 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%view_290, [2, 256, 1024]), kwargs = {})
%view_291 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_17, [2, 256, 1024]), kwargs = {})
%expand_18 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%t_110, [2, 1024, 1024]), kwargs = {})
%view_292 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_18, [2, 1024, 1024]), kwargs = {})
%bmm_24 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%view_291, %view_292), kwargs = {})
%_unsafe_view_120 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%bmm_24, [2, 256, 1024]), kwargs = {})
%_param_constant352 : [#users=1] = get_attr[target=_param_constant352]
%add_179 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%_unsafe_view_120, %_param_constant352), kwargs = {})
%convert_element_type_180 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_179, torch.float32), kwargs = {})
%var_mean_57 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_180, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_114 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_57, 0), kwargs = {})
%getitem_115 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_57, 1), kwargs = {})
%add_180 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_114, 1e-05), kwargs = {})
%rsqrt_57 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_180,), kwargs = {})
%sub_57 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_179, %getitem_115), kwargs = {})
%mul_125 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_57, %rsqrt_57), kwargs = {})
%_param_constant353 : [#users=1] = get_attr[target=_param_constant353]
%mul_126 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_125, %_param_constant353), kwargs = {})
%_param_constant354 : [#users=1] = get_attr[target=_param_constant354]
%add_181 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_126, %_param_constant354), kwargs = {})
%convert_element_type_181 : [#users=3] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_181, torch.float16), kwargs = {})
%_param_constant355 : [#users=1] = get_attr[target=_param_constant355]
%t_111 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant355,), kwargs = {})
%view_293 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_181, [512, 1024]), kwargs = {})
%mm_48 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_293, %t_111), kwargs = {})
%_unsafe_view_121 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_48, [2, 256, 1024]), kwargs = {})
%view_294 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_121, [2, 256, 8, 128]), kwargs = {})
%permute_81 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_294, [0, 2, 1, 3]), kwargs = {})
%clone_72 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_81,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_122 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_72, [16, 256, 128]), kwargs = {})
%_param_constant356 : [#users=1] = get_attr[target=_param_constant356]
%t_112 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant356,), kwargs = {})
%view_295 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_181, [512, 1024]), kwargs = {})
%mm_49 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_295, %t_112), kwargs = {})
%_unsafe_view_123 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_49, [2, 256, 1024]), kwargs = {})
%_param_constant357 : [#users=1] = get_attr[target=_param_constant357]
%t_113 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant357,), kwargs = {})
%view_296 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_181, [512, 1024]), kwargs = {})
%mm_50 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_296, %t_113), kwargs = {})
%_unsafe_view_124 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_50, [2, 256, 1024]), kwargs = {})
%view_297 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_123, [2, 256, 8, 128]), kwargs = {})
%permute_82 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_297, [0, 2, 1, 3]), kwargs = {})
%clone_73 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_82,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_125 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_73, [16, 256, 128]), kwargs = {})
%view_298 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_124, [2, 256, 8, 128]), kwargs = {})
%permute_83 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_298, [0, 2, 1, 3]), kwargs = {})
%clone_74 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_83,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_126 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_74, [16, 256, 128]), kwargs = {})
%empty_16 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 256, 256],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_16 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_125, -1, -2), kwargs = {})
%baddbmm_16 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_16, %_unsafe_view_122, %transpose_16), kwargs = {beta: 0, alpha: 0.08838834764831845})
%_softmax_16 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_16, -1, False), kwargs = {})
%detach_82 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_16,), kwargs = {})
%bmm_25 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_16, %_unsafe_view_126), kwargs = {})
%view_299 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_25, [2, 8, 256, 128]), kwargs = {})
%permute_84 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_299, [0, 2, 1, 3]), kwargs = {})
%clone_75 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_84,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_127 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_75, [2, 256, 1024]), kwargs = {})
%_param_constant358 : [#users=1] = get_attr[target=_param_constant358]
%t_114 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant358,), kwargs = {})
%view_300 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_127, [512, 1024]), kwargs = {})
%_param_constant359 : [#users=1] = get_attr[target=_param_constant359]
%addmm_54 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant359, %view_300, %t_114), kwargs = {})
%view_301 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_54, [2, 256, 1024]), kwargs = {})
%add_182 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_301, %add_179), kwargs = {})
%convert_element_type_182 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_182, torch.float32), kwargs = {})
%var_mean_58 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_182, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_116 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_58, 0), kwargs = {})
%getitem_117 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_58, 1), kwargs = {})
%add_183 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_116, 1e-05), kwargs = {})
%rsqrt_58 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_183,), kwargs = {})
%sub_58 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_182, %getitem_117), kwargs = {})
%mul_127 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_58, %rsqrt_58), kwargs = {})
%_param_constant360 : [#users=1] = get_attr[target=_param_constant360]
%mul_128 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_127, %_param_constant360), kwargs = {})
%_param_constant361 : [#users=1] = get_attr[target=_param_constant361]
%add_184 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_128, %_param_constant361), kwargs = {})
%convert_element_type_183 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_184, torch.float16), kwargs = {})
%_param_constant362 : [#users=1] = get_attr[target=_param_constant362]
%t_115 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant362,), kwargs = {})
%view_302 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_183, [512, 1024]), kwargs = {})
%mm_51 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_302, %t_115), kwargs = {})
%_unsafe_view_128 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_51, [2, 256, 1024]), kwargs = {})
%view_303 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_128, [2, 256, 8, 128]), kwargs = {})
%permute_85 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_303, [0, 2, 1, 3]), kwargs = {})
%clone_76 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_85,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_129 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_76, [16, 256, 128]), kwargs = {})
%_param_constant363 : [#users=1] = get_attr[target=_param_constant363]
%t_116 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant363,), kwargs = {})
%view_304 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_52 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_304, %t_116), kwargs = {})
%_unsafe_view_130 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_52, [2, 77, 1024]), kwargs = {})
%_param_constant364 : [#users=1] = get_attr[target=_param_constant364]
%t_117 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant364,), kwargs = {})
%view_305 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_53 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_305, %t_117), kwargs = {})
%_unsafe_view_131 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_53, [2, 77, 1024]), kwargs = {})
%view_306 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_130, [2, 77, 8, 128]), kwargs = {})
%permute_86 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_306, [0, 2, 1, 3]), kwargs = {})
%clone_77 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_86,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_132 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_77, [16, 77, 128]), kwargs = {})
%view_307 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_131, [2, 77, 8, 128]), kwargs = {})
%permute_87 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_307, [0, 2, 1, 3]), kwargs = {})
%clone_78 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_87,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_133 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_78, [16, 77, 128]), kwargs = {})
%empty_17 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 256, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_17 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_132, -1, -2), kwargs = {})
%baddbmm_17 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_17, %_unsafe_view_129, %transpose_17), kwargs = {beta: 0, alpha: 0.08838834764831845})
%_softmax_17 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_17, -1, False), kwargs = {})
%detach_83 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_17,), kwargs = {})
%bmm_26 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_17, %_unsafe_view_133), kwargs = {})
%view_308 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_26, [2, 8, 256, 128]), kwargs = {})
%permute_88 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_308, [0, 2, 1, 3]), kwargs = {})
%clone_79 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_88,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_134 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_79, [2, 256, 1024]), kwargs = {})
%_param_constant365 : [#users=1] = get_attr[target=_param_constant365]
%t_118 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant365,), kwargs = {})
%view_309 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_134, [512, 1024]), kwargs = {})
%_param_constant366 : [#users=1] = get_attr[target=_param_constant366]
%addmm_55 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant366, %view_309, %t_118), kwargs = {})
%view_310 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_55, [2, 256, 1024]), kwargs = {})
%add_185 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_310, %add_182), kwargs = {})
%convert_element_type_184 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_185, torch.float32), kwargs = {})
%var_mean_59 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_184, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_118 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_59, 0), kwargs = {})
%getitem_119 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_59, 1), kwargs = {})
%add_186 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_118, 1e-05), kwargs = {})
%rsqrt_59 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_186,), kwargs = {})
%sub_59 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_185, %getitem_119), kwargs = {})
%mul_129 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_59, %rsqrt_59), kwargs = {})
%_param_constant367 : [#users=1] = get_attr[target=_param_constant367]
%mul_130 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_129, %_param_constant367), kwargs = {})
%_param_constant368 : [#users=1] = get_attr[target=_param_constant368]
%add_187 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_130, %_param_constant368), kwargs = {})
%convert_element_type_185 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_187, torch.float16), kwargs = {})
%_param_constant369 : [#users=1] = get_attr[target=_param_constant369]
%t_119 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant369,), kwargs = {})
%view_311 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_185, [512, 1024]), kwargs = {})
%_param_constant370 : [#users=1] = get_attr[target=_param_constant370]
%addmm_56 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant370, %view_311, %t_119), kwargs = {})
%view_312 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%addmm_56, [2, 256, 8192]), kwargs = {})
%slice_47 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_312, -1, 0, 4096), kwargs = {})
%slice_48 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_312, -1, 4096, 8192), kwargs = {})
%gelu_8 : [#users=1] = call_function[target=torch.ops.aten.gelu](args = (%slice_48,), kwargs = {})
%mul_131 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%slice_47, %gelu_8), kwargs = {})
%_param_constant371 : [#users=1] = get_attr[target=_param_constant371]
%t_120 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant371,), kwargs = {})
%view_313 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_131, [512, 4096]), kwargs = {})
%_param_constant372 : [#users=1] = get_attr[target=_param_constant372]
%addmm_57 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant372, %view_313, %t_120), kwargs = {})
%view_314 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_57, [2, 256, 1024]), kwargs = {})
%add_188 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%view_314, %add_185), kwargs = {})
%_param_constant373 : [#users=1] = get_attr[target=_param_constant373]
%t_121 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant373,), kwargs = {})
%view_315 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%add_188, [512, 1024]), kwargs = {})
%_param_constant374 : [#users=1] = get_attr[target=_param_constant374]
%addmm_58 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant374, %view_315, %t_121), kwargs = {})
%view_316 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_58, [2, 256, 1024]), kwargs = {})
%view_317 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%view_316, [2, 16, 16, 1024]), kwargs = {})
%permute_89 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_317, [0, 3, 1, 2]), kwargs = {})
%clone_80 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_89,), kwargs = {memory_format: torch.contiguous_format})
%add_189 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%clone_80, %div_12), kwargs = {})
%cat_4 : [#users=2] = call_function[target=torch.ops.aten.cat](args = ([%add_189, %convolution_16], 1), kwargs = {})
%view_318 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%cat_4, [2, 32, 48, 256]), kwargs = {})
%convert_element_type_186 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_318, torch.float32), kwargs = {})
%var_mean_60 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_186, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_120 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_60, 0), kwargs = {})
%getitem_121 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_60, 1), kwargs = {})
%add_190 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_120, 1e-05), kwargs = {})
%rsqrt_60 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_190,), kwargs = {})
%sub_60 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_318, %getitem_121), kwargs = {})
%mul_132 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_60, %rsqrt_60), kwargs = {})
%view_319 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_132, [2, 1536, 16, 16]), kwargs = {})
%_param_constant375 : [#users=1] = get_attr[target=_param_constant375]
%unsqueeze_224 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant375, 0), kwargs = {})
%unsqueeze_225 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_224, 2), kwargs = {})
%unsqueeze_226 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_225, 3), kwargs = {})
%_param_constant376 : [#users=1] = get_attr[target=_param_constant376]
%unsqueeze_227 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant376, 0), kwargs = {})
%unsqueeze_228 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_227, 2), kwargs = {})
%unsqueeze_229 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_228, 3), kwargs = {})
%mul_133 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_319, %unsqueeze_229), kwargs = {})
%add_191 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_133, %unsqueeze_226), kwargs = {})
%convert_element_type_187 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_191, torch.float16), kwargs = {})
%convert_element_type_188 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_121, torch.float16), kwargs = {})
%convert_element_type_189 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_60, torch.float16), kwargs = {})
%squeeze_132 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_188, 3), kwargs = {})
%squeeze_133 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_132, 2), kwargs = {})
%squeeze_134 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_189, 3), kwargs = {})
%squeeze_135 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_134, 2), kwargs = {})
%detach_84 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_133,), kwargs = {})
%detach_85 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_135,), kwargs = {})
%silu_37 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_187,), kwargs = {})
%_param_constant377 : [#users=1] = get_attr[target=_param_constant377]
%_param_constant378 : [#users=1] = get_attr[target=_param_constant378]
%convolution_32 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_37, %_param_constant377, %_param_constant378, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_38 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant379 : [#users=1] = get_attr[target=_param_constant379]
%t_122 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant379,), kwargs = {})
%_param_constant380 : [#users=1] = get_attr[target=_param_constant380]
%addmm_59 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant380, %silu_38, %t_122), kwargs = {})
%slice_49 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_59, 0, 0, 9223372036854775807), kwargs = {})
%slice_50 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_49, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_230 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_50, 2), kwargs = {})
%unsqueeze_231 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_230, 3), kwargs = {})
%add_192 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_32, %unsqueeze_231), kwargs = {})
%view_320 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_192, [2, 32, 32, 256]), kwargs = {})
%convert_element_type_190 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_320, torch.float32), kwargs = {})
%var_mean_61 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_190, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_122 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_61, 0), kwargs = {})
%getitem_123 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_61, 1), kwargs = {})
%add_193 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_122, 1e-05), kwargs = {})
%rsqrt_61 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_193,), kwargs = {})
%sub_61 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_320, %getitem_123), kwargs = {})
%mul_134 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_61, %rsqrt_61), kwargs = {})
%view_321 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_134, [2, 1024, 16, 16]), kwargs = {})
%_param_constant381 : [#users=1] = get_attr[target=_param_constant381]
%unsqueeze_232 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant381, 0), kwargs = {})
%unsqueeze_233 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_232, 2), kwargs = {})
%unsqueeze_234 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_233, 3), kwargs = {})
%_param_constant382 : [#users=1] = get_attr[target=_param_constant382]
%unsqueeze_235 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant382, 0), kwargs = {})
%unsqueeze_236 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_235, 2), kwargs = {})
%unsqueeze_237 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_236, 3), kwargs = {})
%mul_135 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_321, %unsqueeze_237), kwargs = {})
%add_194 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_135, %unsqueeze_234), kwargs = {})
%convert_element_type_191 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_194, torch.float16), kwargs = {})
%convert_element_type_192 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_123, torch.float16), kwargs = {})
%convert_element_type_193 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_61, torch.float16), kwargs = {})
%squeeze_136 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_192, 3), kwargs = {})
%squeeze_137 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_136, 2), kwargs = {})
%squeeze_138 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_193, 3), kwargs = {})
%squeeze_139 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_138, 2), kwargs = {})
%detach_86 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_137,), kwargs = {})
%detach_87 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_139,), kwargs = {})
%silu_39 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_191,), kwargs = {})
%_param_constant383 : [#users=1] = get_attr[target=_param_constant383]
%_param_constant384 : [#users=1] = get_attr[target=_param_constant384]
%convolution_33 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_39, %_param_constant383, %_param_constant384, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%_param_constant385 : [#users=1] = get_attr[target=_param_constant385]
%_param_constant386 : [#users=1] = get_attr[target=_param_constant386]
%convolution_34 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%cat_4, %_param_constant385, %_param_constant386, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
%add_195 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_34, %convolution_33), kwargs = {})
%div_13 : [#users=2] = call_function[target=torch.ops.aten.div](args = (%add_195, 1.0), kwargs = {})
%view_322 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%div_13, [2, 32, 32, 256]), kwargs = {})
%convert_element_type_194 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_322, torch.float32), kwargs = {})
%var_mean_62 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_194, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_124 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_62, 0), kwargs = {})
%getitem_125 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_62, 1), kwargs = {})
%add_196 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_124, 1e-06), kwargs = {})
%rsqrt_62 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_196,), kwargs = {})
%sub_62 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_322, %getitem_125), kwargs = {})
%mul_136 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_62, %rsqrt_62), kwargs = {})
%view_323 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_136, [2, 1024, 16, 16]), kwargs = {})
%_param_constant387 : [#users=1] = get_attr[target=_param_constant387]
%unsqueeze_238 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant387, 0), kwargs = {})
%unsqueeze_239 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_238, 2), kwargs = {})
%unsqueeze_240 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_239, 3), kwargs = {})
%_param_constant388 : [#users=1] = get_attr[target=_param_constant388]
%unsqueeze_241 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant388, 0), kwargs = {})
%unsqueeze_242 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_241, 2), kwargs = {})
%unsqueeze_243 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_242, 3), kwargs = {})
%mul_137 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_323, %unsqueeze_243), kwargs = {})
%add_197 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_137, %unsqueeze_240), kwargs = {})
%convert_element_type_195 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_197, torch.float16), kwargs = {})
%convert_element_type_196 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_125, torch.float16), kwargs = {})
%convert_element_type_197 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_62, torch.float16), kwargs = {})
%squeeze_140 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_196, 3), kwargs = {})
%squeeze_141 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_140, 2), kwargs = {})
%squeeze_142 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_197, 3), kwargs = {})
%squeeze_143 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_142, 2), kwargs = {})
%detach_88 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_141,), kwargs = {})
%detach_89 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_143,), kwargs = {})
%permute_90 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%convert_element_type_195, [0, 2, 3, 1]), kwargs = {})
%view_324 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%permute_90, [2, 256, 1024]), kwargs = {})
%_param_constant389 : [#users=1] = get_attr[target=_param_constant389]
%t_123 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant389,), kwargs = {})
%expand_19 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%view_324, [2, 256, 1024]), kwargs = {})
%view_325 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_19, [2, 256, 1024]), kwargs = {})
%expand_20 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%t_123, [2, 1024, 1024]), kwargs = {})
%view_326 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_20, [2, 1024, 1024]), kwargs = {})
%bmm_27 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%view_325, %view_326), kwargs = {})
%_unsafe_view_135 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%bmm_27, [2, 256, 1024]), kwargs = {})
%_param_constant390 : [#users=1] = get_attr[target=_param_constant390]
%add_198 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%_unsafe_view_135, %_param_constant390), kwargs = {})
%convert_element_type_198 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_198, torch.float32), kwargs = {})
%var_mean_63 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_198, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_126 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_63, 0), kwargs = {})
%getitem_127 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_63, 1), kwargs = {})
%add_199 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_126, 1e-05), kwargs = {})
%rsqrt_63 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_199,), kwargs = {})
%sub_63 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_198, %getitem_127), kwargs = {})
%mul_138 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_63, %rsqrt_63), kwargs = {})
%_param_constant391 : [#users=1] = get_attr[target=_param_constant391]
%mul_139 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_138, %_param_constant391), kwargs = {})
%_param_constant392 : [#users=1] = get_attr[target=_param_constant392]
%add_200 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_139, %_param_constant392), kwargs = {})
%convert_element_type_199 : [#users=3] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_200, torch.float16), kwargs = {})
%_param_constant393 : [#users=1] = get_attr[target=_param_constant393]
%t_124 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant393,), kwargs = {})
%view_327 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_199, [512, 1024]), kwargs = {})
%mm_54 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_327, %t_124), kwargs = {})
%_unsafe_view_136 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_54, [2, 256, 1024]), kwargs = {})
%view_328 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_136, [2, 256, 8, 128]), kwargs = {})
%permute_91 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_328, [0, 2, 1, 3]), kwargs = {})
%clone_81 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_91,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_137 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_81, [16, 256, 128]), kwargs = {})
%_param_constant394 : [#users=1] = get_attr[target=_param_constant394]
%t_125 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant394,), kwargs = {})
%view_329 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_199, [512, 1024]), kwargs = {})
%mm_55 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_329, %t_125), kwargs = {})
%_unsafe_view_138 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_55, [2, 256, 1024]), kwargs = {})
%_param_constant395 : [#users=1] = get_attr[target=_param_constant395]
%t_126 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant395,), kwargs = {})
%view_330 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_199, [512, 1024]), kwargs = {})
%mm_56 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_330, %t_126), kwargs = {})
%_unsafe_view_139 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_56, [2, 256, 1024]), kwargs = {})
%view_331 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_138, [2, 256, 8, 128]), kwargs = {})
%permute_92 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_331, [0, 2, 1, 3]), kwargs = {})
%clone_82 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_92,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_140 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_82, [16, 256, 128]), kwargs = {})
%view_332 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_139, [2, 256, 8, 128]), kwargs = {})
%permute_93 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_332, [0, 2, 1, 3]), kwargs = {})
%clone_83 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_93,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_141 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_83, [16, 256, 128]), kwargs = {})
%empty_18 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 256, 256],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_18 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_140, -1, -2), kwargs = {})
%baddbmm_18 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_18, %_unsafe_view_137, %transpose_18), kwargs = {beta: 0, alpha: 0.08838834764831845})
%_softmax_18 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_18, -1, False), kwargs = {})
%detach_90 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_18,), kwargs = {})
%bmm_28 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_18, %_unsafe_view_141), kwargs = {})
%view_333 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_28, [2, 8, 256, 128]), kwargs = {})
%permute_94 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_333, [0, 2, 1, 3]), kwargs = {})
%clone_84 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_94,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_142 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_84, [2, 256, 1024]), kwargs = {})
%_param_constant396 : [#users=1] = get_attr[target=_param_constant396]
%t_127 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant396,), kwargs = {})
%view_334 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_142, [512, 1024]), kwargs = {})
%_param_constant397 : [#users=1] = get_attr[target=_param_constant397]
%addmm_60 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant397, %view_334, %t_127), kwargs = {})
%view_335 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_60, [2, 256, 1024]), kwargs = {})
%add_201 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_335, %add_198), kwargs = {})
%convert_element_type_200 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_201, torch.float32), kwargs = {})
%var_mean_64 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_200, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_128 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_64, 0), kwargs = {})
%getitem_129 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_64, 1), kwargs = {})
%add_202 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_128, 1e-05), kwargs = {})
%rsqrt_64 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_202,), kwargs = {})
%sub_64 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_201, %getitem_129), kwargs = {})
%mul_140 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_64, %rsqrt_64), kwargs = {})
%_param_constant398 : [#users=1] = get_attr[target=_param_constant398]
%mul_141 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_140, %_param_constant398), kwargs = {})
%_param_constant399 : [#users=1] = get_attr[target=_param_constant399]
%add_203 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_141, %_param_constant399), kwargs = {})
%convert_element_type_201 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_203, torch.float16), kwargs = {})
%_param_constant400 : [#users=1] = get_attr[target=_param_constant400]
%t_128 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant400,), kwargs = {})
%view_336 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_201, [512, 1024]), kwargs = {})
%mm_57 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_336, %t_128), kwargs = {})
%_unsafe_view_143 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_57, [2, 256, 1024]), kwargs = {})
%view_337 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_143, [2, 256, 8, 128]), kwargs = {})
%permute_95 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_337, [0, 2, 1, 3]), kwargs = {})
%clone_85 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_95,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_144 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_85, [16, 256, 128]), kwargs = {})
%_param_constant401 : [#users=1] = get_attr[target=_param_constant401]
%t_129 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant401,), kwargs = {})
%view_338 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_58 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_338, %t_129), kwargs = {})
%_unsafe_view_145 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_58, [2, 77, 1024]), kwargs = {})
%_param_constant402 : [#users=1] = get_attr[target=_param_constant402]
%t_130 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant402,), kwargs = {})
%view_339 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_59 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_339, %t_130), kwargs = {})
%_unsafe_view_146 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_59, [2, 77, 1024]), kwargs = {})
%view_340 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_145, [2, 77, 8, 128]), kwargs = {})
%permute_96 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_340, [0, 2, 1, 3]), kwargs = {})
%clone_86 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_96,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_147 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_86, [16, 77, 128]), kwargs = {})
%view_341 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_146, [2, 77, 8, 128]), kwargs = {})
%permute_97 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_341, [0, 2, 1, 3]), kwargs = {})
%clone_87 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_97,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_148 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_87, [16, 77, 128]), kwargs = {})
%empty_19 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 256, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_19 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_147, -1, -2), kwargs = {})
%baddbmm_19 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_19, %_unsafe_view_144, %transpose_19), kwargs = {beta: 0, alpha: 0.08838834764831845})
%_softmax_19 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_19, -1, False), kwargs = {})
%detach_91 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_19,), kwargs = {})
%bmm_29 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_19, %_unsafe_view_148), kwargs = {})
%view_342 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_29, [2, 8, 256, 128]), kwargs = {})
%permute_98 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_342, [0, 2, 1, 3]), kwargs = {})
%clone_88 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_98,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_149 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_88, [2, 256, 1024]), kwargs = {})
%_param_constant403 : [#users=1] = get_attr[target=_param_constant403]
%t_131 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant403,), kwargs = {})
%view_343 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_149, [512, 1024]), kwargs = {})
%_param_constant404 : [#users=1] = get_attr[target=_param_constant404]
%addmm_61 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant404, %view_343, %t_131), kwargs = {})
%view_344 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_61, [2, 256, 1024]), kwargs = {})
%add_204 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_344, %add_201), kwargs = {})
%convert_element_type_202 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_204, torch.float32), kwargs = {})
%var_mean_65 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_202, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_130 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_65, 0), kwargs = {})
%getitem_131 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_65, 1), kwargs = {})
%add_205 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_130, 1e-05), kwargs = {})
%rsqrt_65 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_205,), kwargs = {})
%sub_65 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_204, %getitem_131), kwargs = {})
%mul_142 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_65, %rsqrt_65), kwargs = {})
%_param_constant405 : [#users=1] = get_attr[target=_param_constant405]
%mul_143 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_142, %_param_constant405), kwargs = {})
%_param_constant406 : [#users=1] = get_attr[target=_param_constant406]
%add_206 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_143, %_param_constant406), kwargs = {})
%convert_element_type_203 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_206, torch.float16), kwargs = {})
%_param_constant407 : [#users=1] = get_attr[target=_param_constant407]
%t_132 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant407,), kwargs = {})
%view_345 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_203, [512, 1024]), kwargs = {})
%_param_constant408 : [#users=1] = get_attr[target=_param_constant408]
%addmm_62 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant408, %view_345, %t_132), kwargs = {})
%view_346 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%addmm_62, [2, 256, 8192]), kwargs = {})
%slice_51 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_346, -1, 0, 4096), kwargs = {})
%slice_52 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_346, -1, 4096, 8192), kwargs = {})
%gelu_9 : [#users=1] = call_function[target=torch.ops.aten.gelu](args = (%slice_52,), kwargs = {})
%mul_144 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%slice_51, %gelu_9), kwargs = {})
%_param_constant409 : [#users=1] = get_attr[target=_param_constant409]
%t_133 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant409,), kwargs = {})
%view_347 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_144, [512, 4096]), kwargs = {})
%_param_constant410 : [#users=1] = get_attr[target=_param_constant410]
%addmm_63 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant410, %view_347, %t_133), kwargs = {})
%view_348 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_63, [2, 256, 1024]), kwargs = {})
%add_207 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%view_348, %add_204), kwargs = {})
%_param_constant411 : [#users=1] = get_attr[target=_param_constant411]
%t_134 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant411,), kwargs = {})
%view_349 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%add_207, [512, 1024]), kwargs = {})
%_param_constant412 : [#users=1] = get_attr[target=_param_constant412]
%addmm_64 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant412, %view_349, %t_134), kwargs = {})
%view_350 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_64, [2, 256, 1024]), kwargs = {})
%view_351 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%view_350, [2, 16, 16, 1024]), kwargs = {})
%permute_99 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_351, [0, 3, 1, 2]), kwargs = {})
%clone_89 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_99,), kwargs = {memory_format: torch.contiguous_format})
%add_208 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%clone_89, %div_13), kwargs = {})
%upsample_nearest2d : [#users=1] = call_function[target=torch.ops.aten.upsample_nearest2d](args = (%add_208, [32, 32], 2.0, 2.0), kwargs = {})
%_param_constant413 : [#users=1] = get_attr[target=_param_constant413]
%_param_constant414 : [#users=1] = get_attr[target=_param_constant414]
%convolution_35 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%upsample_nearest2d, %_param_constant413, %_param_constant414, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%cat_5 : [#users=2] = call_function[target=torch.ops.aten.cat](args = ([%convolution_35, %add_88], 1), kwargs = {})
%view_352 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%cat_5, [2, 32, 48, 1024]), kwargs = {})
%convert_element_type_204 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_352, torch.float32), kwargs = {})
%var_mean_66 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_204, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_132 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_66, 0), kwargs = {})
%getitem_133 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_66, 1), kwargs = {})
%add_209 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_132, 1e-05), kwargs = {})
%rsqrt_66 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_209,), kwargs = {})
%sub_66 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_352, %getitem_133), kwargs = {})
%mul_145 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_66, %rsqrt_66), kwargs = {})
%view_353 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_145, [2, 1536, 32, 32]), kwargs = {})
%_param_constant415 : [#users=1] = get_attr[target=_param_constant415]
%unsqueeze_244 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant415, 0), kwargs = {})
%unsqueeze_245 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_244, 2), kwargs = {})
%unsqueeze_246 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_245, 3), kwargs = {})
%_param_constant416 : [#users=1] = get_attr[target=_param_constant416]
%unsqueeze_247 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant416, 0), kwargs = {})
%unsqueeze_248 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_247, 2), kwargs = {})
%unsqueeze_249 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_248, 3), kwargs = {})
%mul_146 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_353, %unsqueeze_249), kwargs = {})
%add_210 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_146, %unsqueeze_246), kwargs = {})
%convert_element_type_205 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_210, torch.float16), kwargs = {})
%convert_element_type_206 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_133, torch.float16), kwargs = {})
%convert_element_type_207 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_66, torch.float16), kwargs = {})
%squeeze_144 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_206, 3), kwargs = {})
%squeeze_145 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_144, 2), kwargs = {})
%squeeze_146 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_207, 3), kwargs = {})
%squeeze_147 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_146, 2), kwargs = {})
%detach_92 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_145,), kwargs = {})
%detach_93 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_147,), kwargs = {})
%silu_40 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_205,), kwargs = {})
%_param_constant417 : [#users=1] = get_attr[target=_param_constant417]
%_param_constant418 : [#users=1] = get_attr[target=_param_constant418]
%convolution_36 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_40, %_param_constant417, %_param_constant418, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_41 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant419 : [#users=1] = get_attr[target=_param_constant419]
%t_135 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant419,), kwargs = {})
%_param_constant420 : [#users=1] = get_attr[target=_param_constant420]
%addmm_65 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant420, %silu_41, %t_135), kwargs = {})
%slice_53 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_65, 0, 0, 9223372036854775807), kwargs = {})
%slice_54 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_53, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_250 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_54, 2), kwargs = {})
%unsqueeze_251 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_250, 3), kwargs = {})
%add_211 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_36, %unsqueeze_251), kwargs = {})
%view_354 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_211, [2, 32, 16, 1024]), kwargs = {})
%convert_element_type_208 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_354, torch.float32), kwargs = {})
%var_mean_67 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_208, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_134 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_67, 0), kwargs = {})
%getitem_135 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_67, 1), kwargs = {})
%add_212 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_134, 1e-05), kwargs = {})
%rsqrt_67 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_212,), kwargs = {})
%sub_67 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_354, %getitem_135), kwargs = {})
%mul_147 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_67, %rsqrt_67), kwargs = {})
%view_355 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_147, [2, 512, 32, 32]), kwargs = {})
%_param_constant421 : [#users=1] = get_attr[target=_param_constant421]
%unsqueeze_252 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant421, 0), kwargs = {})
%unsqueeze_253 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_252, 2), kwargs = {})
%unsqueeze_254 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_253, 3), kwargs = {})
%_param_constant422 : [#users=1] = get_attr[target=_param_constant422]
%unsqueeze_255 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant422, 0), kwargs = {})
%unsqueeze_256 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_255, 2), kwargs = {})
%unsqueeze_257 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_256, 3), kwargs = {})
%mul_148 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_355, %unsqueeze_257), kwargs = {})
%add_213 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_148, %unsqueeze_254), kwargs = {})
%convert_element_type_209 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_213, torch.float16), kwargs = {})
%convert_element_type_210 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_135, torch.float16), kwargs = {})
%convert_element_type_211 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_67, torch.float16), kwargs = {})
%squeeze_148 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_210, 3), kwargs = {})
%squeeze_149 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_148, 2), kwargs = {})
%squeeze_150 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_211, 3), kwargs = {})
%squeeze_151 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_150, 2), kwargs = {})
%detach_94 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_149,), kwargs = {})
%detach_95 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_151,), kwargs = {})
%silu_42 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_209,), kwargs = {})
%_param_constant423 : [#users=1] = get_attr[target=_param_constant423]
%_param_constant424 : [#users=1] = get_attr[target=_param_constant424]
%convolution_37 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_42, %_param_constant423, %_param_constant424, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%_param_constant425 : [#users=1] = get_attr[target=_param_constant425]
%_param_constant426 : [#users=1] = get_attr[target=_param_constant426]
%convolution_38 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%cat_5, %_param_constant425, %_param_constant426, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
%add_214 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_38, %convolution_37), kwargs = {})
%div_14 : [#users=2] = call_function[target=torch.ops.aten.div](args = (%add_214, 1.0), kwargs = {})
%view_356 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%div_14, [2, 32, 16, 1024]), kwargs = {})
%convert_element_type_212 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_356, torch.float32), kwargs = {})
%var_mean_68 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_212, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_136 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_68, 0), kwargs = {})
%getitem_137 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_68, 1), kwargs = {})
%add_215 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_136, 1e-06), kwargs = {})
%rsqrt_68 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_215,), kwargs = {})
%sub_68 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_356, %getitem_137), kwargs = {})
%mul_149 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_68, %rsqrt_68), kwargs = {})
%view_357 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_149, [2, 512, 32, 32]), kwargs = {})
%_param_constant427 : [#users=1] = get_attr[target=_param_constant427]
%unsqueeze_258 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant427, 0), kwargs = {})
%unsqueeze_259 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_258, 2), kwargs = {})
%unsqueeze_260 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_259, 3), kwargs = {})
%_param_constant428 : [#users=1] = get_attr[target=_param_constant428]
%unsqueeze_261 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant428, 0), kwargs = {})
%unsqueeze_262 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_261, 2), kwargs = {})
%unsqueeze_263 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_262, 3), kwargs = {})
%mul_150 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_357, %unsqueeze_263), kwargs = {})
%add_216 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_150, %unsqueeze_260), kwargs = {})
%convert_element_type_213 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_216, torch.float16), kwargs = {})
%convert_element_type_214 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_137, torch.float16), kwargs = {})
%convert_element_type_215 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_68, torch.float16), kwargs = {})
%squeeze_152 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_214, 3), kwargs = {})
%squeeze_153 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_152, 2), kwargs = {})
%squeeze_154 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_215, 3), kwargs = {})
%squeeze_155 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_154, 2), kwargs = {})
%detach_96 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_153,), kwargs = {})
%detach_97 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_155,), kwargs = {})
%permute_100 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%convert_element_type_213, [0, 2, 3, 1]), kwargs = {})
%view_358 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%permute_100, [2, 1024, 512]), kwargs = {})
%_param_constant429 : [#users=1] = get_attr[target=_param_constant429]
%t_136 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant429,), kwargs = {})
%expand_21 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%view_358, [2, 1024, 512]), kwargs = {})
%view_359 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_21, [2, 1024, 512]), kwargs = {})
%expand_22 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%t_136, [2, 512, 512]), kwargs = {})
%view_360 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_22, [2, 512, 512]), kwargs = {})
%bmm_30 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%view_359, %view_360), kwargs = {})
%_unsafe_view_150 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%bmm_30, [2, 1024, 512]), kwargs = {})
%_param_constant430 : [#users=1] = get_attr[target=_param_constant430]
%add_217 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%_unsafe_view_150, %_param_constant430), kwargs = {})
%convert_element_type_216 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_217, torch.float32), kwargs = {})
%var_mean_69 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_216, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_138 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_69, 0), kwargs = {})
%getitem_139 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_69, 1), kwargs = {})
%add_218 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_138, 1e-05), kwargs = {})
%rsqrt_69 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_218,), kwargs = {})
%sub_69 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_217, %getitem_139), kwargs = {})
%mul_151 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_69, %rsqrt_69), kwargs = {})
%_param_constant431 : [#users=1] = get_attr[target=_param_constant431]
%mul_152 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_151, %_param_constant431), kwargs = {})
%_param_constant432 : [#users=1] = get_attr[target=_param_constant432]
%add_219 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_152, %_param_constant432), kwargs = {})
%convert_element_type_217 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_219, torch.float16), kwargs = {})
%_param_constant433 : [#users=1] = get_attr[target=_param_constant433]
%t_137 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant433,), kwargs = {})
%view_361 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_217, [2048, 512]), kwargs = {})
%mm_60 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_361, %t_137), kwargs = {})
%_unsafe_view_151 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_60, [2, 1024, 512]), kwargs = {})
%view_362 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_151, [2, 1024, 8, 64]), kwargs = {})
%permute_101 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_362, [0, 2, 1, 3]), kwargs = {})
%clone_90 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_101,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_152 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_90, [16, 1024, 64]), kwargs = {})
%_param_constant434 : [#users=1] = get_attr[target=_param_constant434]
%t_138 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant434,), kwargs = {})
%view_363 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_61 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_363, %t_138), kwargs = {})
%_unsafe_view_153 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_61, [2, 77, 512]), kwargs = {})
%_param_constant435 : [#users=1] = get_attr[target=_param_constant435]
%t_139 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant435,), kwargs = {})
%view_364 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_62 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_364, %t_139), kwargs = {})
%_unsafe_view_154 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_62, [2, 77, 512]), kwargs = {})
%view_365 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_153, [2, 77, 8, 64]), kwargs = {})
%permute_102 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_365, [0, 2, 1, 3]), kwargs = {})
%clone_91 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_102,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_155 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_91, [16, 77, 64]), kwargs = {})
%view_366 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_154, [2, 77, 8, 64]), kwargs = {})
%permute_103 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_366, [0, 2, 1, 3]), kwargs = {})
%clone_92 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_103,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_156 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_92, [16, 77, 64]), kwargs = {})
%empty_20 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 1024, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_20 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_155, -1, -2), kwargs = {})
%baddbmm_20 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_20, %_unsafe_view_152, %transpose_20), kwargs = {beta: 0, alpha: 0.125})
%_softmax_20 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_20, -1, False), kwargs = {})
%detach_98 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_20,), kwargs = {})
%bmm_31 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_20, %_unsafe_view_156), kwargs = {})
%view_367 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_31, [2, 8, 1024, 64]), kwargs = {})
%permute_104 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_367, [0, 2, 1, 3]), kwargs = {})
%clone_93 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_104,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_157 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_93, [2, 1024, 512]), kwargs = {})
%_param_constant436 : [#users=1] = get_attr[target=_param_constant436]
%t_140 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant436,), kwargs = {})
%view_368 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_157, [2048, 512]), kwargs = {})
%_param_constant437 : [#users=1] = get_attr[target=_param_constant437]
%addmm_66 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant437, %view_368, %t_140), kwargs = {})
%view_369 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_66, [2, 1024, 512]), kwargs = {})
%add_220 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_369, %add_217), kwargs = {})
%convert_element_type_218 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_220, torch.float32), kwargs = {})
%var_mean_70 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_218, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_140 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_70, 0), kwargs = {})
%getitem_141 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_70, 1), kwargs = {})
%add_221 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_140, 1e-05), kwargs = {})
%rsqrt_70 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_221,), kwargs = {})
%sub_70 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_220, %getitem_141), kwargs = {})
%mul_153 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_70, %rsqrt_70), kwargs = {})
%_param_constant438 : [#users=1] = get_attr[target=_param_constant438]
%mul_154 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_153, %_param_constant438), kwargs = {})
%_param_constant439 : [#users=1] = get_attr[target=_param_constant439]
%add_222 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_154, %_param_constant439), kwargs = {})
%convert_element_type_219 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_222, torch.float16), kwargs = {})
%_param_constant440 : [#users=1] = get_attr[target=_param_constant440]
%t_141 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant440,), kwargs = {})
%view_370 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_219, [2048, 512]), kwargs = {})
%mm_63 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_370, %t_141), kwargs = {})
%_unsafe_view_158 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_63, [2, 1024, 512]), kwargs = {})
%view_371 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_158, [2, 1024, 8, 64]), kwargs = {})
%permute_105 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_371, [0, 2, 1, 3]), kwargs = {})
%clone_94 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_105,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_159 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_94, [16, 1024, 64]), kwargs = {})
%_param_constant441 : [#users=1] = get_attr[target=_param_constant441]
%t_142 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant441,), kwargs = {})
%view_372 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_64 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_372, %t_142), kwargs = {})
%_unsafe_view_160 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_64, [2, 77, 512]), kwargs = {})
%_param_constant442 : [#users=1] = get_attr[target=_param_constant442]
%t_143 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant442,), kwargs = {})
%view_373 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_65 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_373, %t_143), kwargs = {})
%_unsafe_view_161 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_65, [2, 77, 512]), kwargs = {})
%view_374 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_160, [2, 77, 8, 64]), kwargs = {})
%permute_106 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_374, [0, 2, 1, 3]), kwargs = {})
%clone_95 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_106,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_162 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_95, [16, 77, 64]), kwargs = {})
%view_375 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_161, [2, 77, 8, 64]), kwargs = {})
%permute_107 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_375, [0, 2, 1, 3]), kwargs = {})
%clone_96 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_107,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_163 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_96, [16, 77, 64]), kwargs = {})
%empty_21 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 1024, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_21 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_162, -1, -2), kwargs = {})
%baddbmm_21 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_21, %_unsafe_view_159, %transpose_21), kwargs = {beta: 0, alpha: 0.125})
%_softmax_21 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_21, -1, False), kwargs = {})
%detach_99 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_21,), kwargs = {})
%bmm_32 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_21, %_unsafe_view_163), kwargs = {})
%view_376 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_32, [2, 8, 1024, 64]), kwargs = {})
%permute_108 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_376, [0, 2, 1, 3]), kwargs = {})
%clone_97 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_108,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_164 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_97, [2, 1024, 512]), kwargs = {})
%_param_constant443 : [#users=1] = get_attr[target=_param_constant443]
%t_144 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant443,), kwargs = {})
%view_377 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_164, [2048, 512]), kwargs = {})
%_param_constant444 : [#users=1] = get_attr[target=_param_constant444]
%addmm_67 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant444, %view_377, %t_144), kwargs = {})
%view_378 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_67, [2, 1024, 512]), kwargs = {})
%add_223 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_378, %add_220), kwargs = {})
%convert_element_type_220 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_223, torch.float32), kwargs = {})
%var_mean_71 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_220, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_142 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_71, 0), kwargs = {})
%getitem_143 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_71, 1), kwargs = {})
%add_224 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_142, 1e-05), kwargs = {})
%rsqrt_71 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_224,), kwargs = {})
%sub_71 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_223, %getitem_143), kwargs = {})
%mul_155 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_71, %rsqrt_71), kwargs = {})
%_param_constant445 : [#users=1] = get_attr[target=_param_constant445]
%mul_156 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_155, %_param_constant445), kwargs = {})
%_param_constant446 : [#users=1] = get_attr[target=_param_constant446]
%add_225 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_156, %_param_constant446), kwargs = {})
%convert_element_type_221 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_225, torch.float16), kwargs = {})
%_param_constant447 : [#users=1] = get_attr[target=_param_constant447]
%t_145 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant447,), kwargs = {})
%view_379 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_221, [2048, 512]), kwargs = {})
%_param_constant448 : [#users=1] = get_attr[target=_param_constant448]
%addmm_68 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant448, %view_379, %t_145), kwargs = {})
%view_380 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%addmm_68, [2, 1024, 4096]), kwargs = {})
%slice_55 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_380, -1, 0, 2048), kwargs = {})
%slice_56 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_380, -1, 2048, 4096), kwargs = {})
%gelu_10 : [#users=1] = call_function[target=torch.ops.aten.gelu](args = (%slice_56,), kwargs = {})
%mul_157 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%slice_55, %gelu_10), kwargs = {})
%_param_constant449 : [#users=1] = get_attr[target=_param_constant449]
%t_146 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant449,), kwargs = {})
%view_381 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_157, [2048, 2048]), kwargs = {})
%_param_constant450 : [#users=1] = get_attr[target=_param_constant450]
%addmm_69 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant450, %view_381, %t_146), kwargs = {})
%view_382 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_69, [2, 1024, 512]), kwargs = {})
%add_226 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%view_382, %add_223), kwargs = {})
%_param_constant451 : [#users=1] = get_attr[target=_param_constant451]
%t_147 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant451,), kwargs = {})
%view_383 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%add_226, [2048, 512]), kwargs = {})
%_param_constant452 : [#users=1] = get_attr[target=_param_constant452]
%addmm_70 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant452, %view_383, %t_147), kwargs = {})
%view_384 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_70, [2, 1024, 512]), kwargs = {})
%view_385 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%view_384, [2, 32, 32, 512]), kwargs = {})
%permute_109 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_385, [0, 3, 1, 2]), kwargs = {})
%clone_98 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_109,), kwargs = {memory_format: torch.contiguous_format})
%add_227 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%clone_98, %div_14), kwargs = {})
%cat_6 : [#users=2] = call_function[target=torch.ops.aten.cat](args = ([%add_227, %add_69], 1), kwargs = {})
%view_386 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%cat_6, [2, 32, 32, 1024]), kwargs = {})
%convert_element_type_222 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_386, torch.float32), kwargs = {})
%var_mean_72 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_222, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_144 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_72, 0), kwargs = {})
%getitem_145 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_72, 1), kwargs = {})
%add_228 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_144, 1e-05), kwargs = {})
%rsqrt_72 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_228,), kwargs = {})
%sub_72 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_386, %getitem_145), kwargs = {})
%mul_158 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_72, %rsqrt_72), kwargs = {})
%view_387 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_158, [2, 1024, 32, 32]), kwargs = {})
%_param_constant453 : [#users=1] = get_attr[target=_param_constant453]
%unsqueeze_264 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant453, 0), kwargs = {})
%unsqueeze_265 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_264, 2), kwargs = {})
%unsqueeze_266 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_265, 3), kwargs = {})
%_param_constant454 : [#users=1] = get_attr[target=_param_constant454]
%unsqueeze_267 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant454, 0), kwargs = {})
%unsqueeze_268 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_267, 2), kwargs = {})
%unsqueeze_269 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_268, 3), kwargs = {})
%mul_159 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_387, %unsqueeze_269), kwargs = {})
%add_229 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_159, %unsqueeze_266), kwargs = {})
%convert_element_type_223 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_229, torch.float16), kwargs = {})
%convert_element_type_224 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_145, torch.float16), kwargs = {})
%convert_element_type_225 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_72, torch.float16), kwargs = {})
%squeeze_156 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_224, 3), kwargs = {})
%squeeze_157 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_156, 2), kwargs = {})
%squeeze_158 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_225, 3), kwargs = {})
%squeeze_159 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_158, 2), kwargs = {})
%detach_100 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_157,), kwargs = {})
%detach_101 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_159,), kwargs = {})
%silu_43 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_223,), kwargs = {})
%_param_constant455 : [#users=1] = get_attr[target=_param_constant455]
%_param_constant456 : [#users=1] = get_attr[target=_param_constant456]
%convolution_39 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_43, %_param_constant455, %_param_constant456, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_44 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant457 : [#users=1] = get_attr[target=_param_constant457]
%t_148 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant457,), kwargs = {})
%_param_constant458 : [#users=1] = get_attr[target=_param_constant458]
%addmm_71 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant458, %silu_44, %t_148), kwargs = {})
%slice_57 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_71, 0, 0, 9223372036854775807), kwargs = {})
%slice_58 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_57, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_270 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_58, 2), kwargs = {})
%unsqueeze_271 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_270, 3), kwargs = {})
%add_230 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_39, %unsqueeze_271), kwargs = {})
%view_388 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_230, [2, 32, 16, 1024]), kwargs = {})
%convert_element_type_226 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_388, torch.float32), kwargs = {})
%var_mean_73 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_226, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_146 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_73, 0), kwargs = {})
%getitem_147 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_73, 1), kwargs = {})
%add_231 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_146, 1e-05), kwargs = {})
%rsqrt_73 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_231,), kwargs = {})
%sub_73 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_388, %getitem_147), kwargs = {})
%mul_160 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_73, %rsqrt_73), kwargs = {})
%view_389 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_160, [2, 512, 32, 32]), kwargs = {})
%_param_constant459 : [#users=1] = get_attr[target=_param_constant459]
%unsqueeze_272 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant459, 0), kwargs = {})
%unsqueeze_273 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_272, 2), kwargs = {})
%unsqueeze_274 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_273, 3), kwargs = {})
%_param_constant460 : [#users=1] = get_attr[target=_param_constant460]
%unsqueeze_275 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant460, 0), kwargs = {})
%unsqueeze_276 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_275, 2), kwargs = {})
%unsqueeze_277 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_276, 3), kwargs = {})
%mul_161 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_389, %unsqueeze_277), kwargs = {})
%add_232 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_161, %unsqueeze_274), kwargs = {})
%convert_element_type_227 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_232, torch.float16), kwargs = {})
%convert_element_type_228 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_147, torch.float16), kwargs = {})
%convert_element_type_229 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_73, torch.float16), kwargs = {})
%squeeze_160 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_228, 3), kwargs = {})
%squeeze_161 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_160, 2), kwargs = {})
%squeeze_162 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_229, 3), kwargs = {})
%squeeze_163 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_162, 2), kwargs = {})
%detach_102 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_161,), kwargs = {})
%detach_103 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_163,), kwargs = {})
%silu_45 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_227,), kwargs = {})
%_param_constant461 : [#users=1] = get_attr[target=_param_constant461]
%_param_constant462 : [#users=1] = get_attr[target=_param_constant462]
%convolution_40 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_45, %_param_constant461, %_param_constant462, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%_param_constant463 : [#users=1] = get_attr[target=_param_constant463]
%_param_constant464 : [#users=1] = get_attr[target=_param_constant464]
%convolution_41 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%cat_6, %_param_constant463, %_param_constant464, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
%add_233 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_41, %convolution_40), kwargs = {})
%div_15 : [#users=2] = call_function[target=torch.ops.aten.div](args = (%add_233, 1.0), kwargs = {})
%view_390 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%div_15, [2, 32, 16, 1024]), kwargs = {})
%convert_element_type_230 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_390, torch.float32), kwargs = {})
%var_mean_74 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_230, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_148 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_74, 0), kwargs = {})
%getitem_149 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_74, 1), kwargs = {})
%add_234 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_148, 1e-06), kwargs = {})
%rsqrt_74 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_234,), kwargs = {})
%sub_74 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_390, %getitem_149), kwargs = {})
%mul_162 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_74, %rsqrt_74), kwargs = {})
%view_391 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_162, [2, 512, 32, 32]), kwargs = {})
%_param_constant465 : [#users=1] = get_attr[target=_param_constant465]
%unsqueeze_278 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant465, 0), kwargs = {})
%unsqueeze_279 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_278, 2), kwargs = {})
%unsqueeze_280 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_279, 3), kwargs = {})
%_param_constant466 : [#users=1] = get_attr[target=_param_constant466]
%unsqueeze_281 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant466, 0), kwargs = {})
%unsqueeze_282 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_281, 2), kwargs = {})
%unsqueeze_283 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_282, 3), kwargs = {})
%mul_163 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_391, %unsqueeze_283), kwargs = {})
%add_235 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_163, %unsqueeze_280), kwargs = {})
%convert_element_type_231 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_235, torch.float16), kwargs = {})
%convert_element_type_232 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_149, torch.float16), kwargs = {})
%convert_element_type_233 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_74, torch.float16), kwargs = {})
%squeeze_164 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_232, 3), kwargs = {})
%squeeze_165 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_164, 2), kwargs = {})
%squeeze_166 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_233, 3), kwargs = {})
%squeeze_167 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_166, 2), kwargs = {})
%detach_104 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_165,), kwargs = {})
%detach_105 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_167,), kwargs = {})
%permute_110 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%convert_element_type_231, [0, 2, 3, 1]), kwargs = {})
%view_392 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%permute_110, [2, 1024, 512]), kwargs = {})
%_param_constant467 : [#users=1] = get_attr[target=_param_constant467]
%t_149 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant467,), kwargs = {})
%expand_23 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%view_392, [2, 1024, 512]), kwargs = {})
%view_393 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_23, [2, 1024, 512]), kwargs = {})
%expand_24 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%t_149, [2, 512, 512]), kwargs = {})
%view_394 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_24, [2, 512, 512]), kwargs = {})
%bmm_33 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%view_393, %view_394), kwargs = {})
%_unsafe_view_165 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%bmm_33, [2, 1024, 512]), kwargs = {})
%_param_constant468 : [#users=1] = get_attr[target=_param_constant468]
%add_236 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%_unsafe_view_165, %_param_constant468), kwargs = {})
%convert_element_type_234 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_236, torch.float32), kwargs = {})
%var_mean_75 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_234, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_150 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_75, 0), kwargs = {})
%getitem_151 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_75, 1), kwargs = {})
%add_237 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_150, 1e-05), kwargs = {})
%rsqrt_75 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_237,), kwargs = {})
%sub_75 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_236, %getitem_151), kwargs = {})
%mul_164 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_75, %rsqrt_75), kwargs = {})
%_param_constant469 : [#users=1] = get_attr[target=_param_constant469]
%mul_165 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_164, %_param_constant469), kwargs = {})
%_param_constant470 : [#users=1] = get_attr[target=_param_constant470]
%add_238 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_165, %_param_constant470), kwargs = {})
%convert_element_type_235 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_238, torch.float16), kwargs = {})
%_param_constant471 : [#users=1] = get_attr[target=_param_constant471]
%t_150 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant471,), kwargs = {})
%view_395 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_235, [2048, 512]), kwargs = {})
%mm_66 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_395, %t_150), kwargs = {})
%_unsafe_view_166 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_66, [2, 1024, 512]), kwargs = {})
%view_396 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_166, [2, 1024, 8, 64]), kwargs = {})
%permute_111 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_396, [0, 2, 1, 3]), kwargs = {})
%clone_99 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_111,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_167 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_99, [16, 1024, 64]), kwargs = {})
%_param_constant472 : [#users=1] = get_attr[target=_param_constant472]
%t_151 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant472,), kwargs = {})
%view_397 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_67 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_397, %t_151), kwargs = {})
%_unsafe_view_168 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_67, [2, 77, 512]), kwargs = {})
%_param_constant473 : [#users=1] = get_attr[target=_param_constant473]
%t_152 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant473,), kwargs = {})
%view_398 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_68 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_398, %t_152), kwargs = {})
%_unsafe_view_169 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_68, [2, 77, 512]), kwargs = {})
%view_399 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_168, [2, 77, 8, 64]), kwargs = {})
%permute_112 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_399, [0, 2, 1, 3]), kwargs = {})
%clone_100 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_112,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_170 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_100, [16, 77, 64]), kwargs = {})
%view_400 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_169, [2, 77, 8, 64]), kwargs = {})
%permute_113 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_400, [0, 2, 1, 3]), kwargs = {})
%clone_101 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_113,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_171 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_101, [16, 77, 64]), kwargs = {})
%empty_22 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 1024, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_22 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_170, -1, -2), kwargs = {})
%baddbmm_22 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_22, %_unsafe_view_167, %transpose_22), kwargs = {beta: 0, alpha: 0.125})
%_softmax_22 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_22, -1, False), kwargs = {})
%detach_106 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_22,), kwargs = {})
%bmm_34 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_22, %_unsafe_view_171), kwargs = {})
%view_401 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_34, [2, 8, 1024, 64]), kwargs = {})
%permute_114 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_401, [0, 2, 1, 3]), kwargs = {})
%clone_102 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_114,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_172 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_102, [2, 1024, 512]), kwargs = {})
%_param_constant474 : [#users=1] = get_attr[target=_param_constant474]
%t_153 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant474,), kwargs = {})
%view_402 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_172, [2048, 512]), kwargs = {})
%_param_constant475 : [#users=1] = get_attr[target=_param_constant475]
%addmm_72 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant475, %view_402, %t_153), kwargs = {})
%view_403 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_72, [2, 1024, 512]), kwargs = {})
%add_239 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_403, %add_236), kwargs = {})
%convert_element_type_236 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_239, torch.float32), kwargs = {})
%var_mean_76 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_236, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_152 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_76, 0), kwargs = {})
%getitem_153 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_76, 1), kwargs = {})
%add_240 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_152, 1e-05), kwargs = {})
%rsqrt_76 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_240,), kwargs = {})
%sub_76 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_239, %getitem_153), kwargs = {})
%mul_166 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_76, %rsqrt_76), kwargs = {})
%_param_constant476 : [#users=1] = get_attr[target=_param_constant476]
%mul_167 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_166, %_param_constant476), kwargs = {})
%_param_constant477 : [#users=1] = get_attr[target=_param_constant477]
%add_241 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_167, %_param_constant477), kwargs = {})
%convert_element_type_237 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_241, torch.float16), kwargs = {})
%_param_constant478 : [#users=1] = get_attr[target=_param_constant478]
%t_154 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant478,), kwargs = {})
%view_404 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_237, [2048, 512]), kwargs = {})
%mm_69 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_404, %t_154), kwargs = {})
%_unsafe_view_173 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_69, [2, 1024, 512]), kwargs = {})
%view_405 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_173, [2, 1024, 8, 64]), kwargs = {})
%permute_115 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_405, [0, 2, 1, 3]), kwargs = {})
%clone_103 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_115,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_174 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_103, [16, 1024, 64]), kwargs = {})
%_param_constant479 : [#users=1] = get_attr[target=_param_constant479]
%t_155 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant479,), kwargs = {})
%view_406 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_70 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_406, %t_155), kwargs = {})
%_unsafe_view_175 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_70, [2, 77, 512]), kwargs = {})
%_param_constant480 : [#users=1] = get_attr[target=_param_constant480]
%t_156 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant480,), kwargs = {})
%view_407 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_71 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_407, %t_156), kwargs = {})
%_unsafe_view_176 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_71, [2, 77, 512]), kwargs = {})
%view_408 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_175, [2, 77, 8, 64]), kwargs = {})
%permute_116 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_408, [0, 2, 1, 3]), kwargs = {})
%clone_104 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_116,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_177 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_104, [16, 77, 64]), kwargs = {})
%view_409 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_176, [2, 77, 8, 64]), kwargs = {})
%permute_117 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_409, [0, 2, 1, 3]), kwargs = {})
%clone_105 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_117,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_178 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_105, [16, 77, 64]), kwargs = {})
%empty_23 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 1024, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_23 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_177, -1, -2), kwargs = {})
%baddbmm_23 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_23, %_unsafe_view_174, %transpose_23), kwargs = {beta: 0, alpha: 0.125})
%_softmax_23 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_23, -1, False), kwargs = {})
%detach_107 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_23,), kwargs = {})
%bmm_35 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_23, %_unsafe_view_178), kwargs = {})
%view_410 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_35, [2, 8, 1024, 64]), kwargs = {})
%permute_118 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_410, [0, 2, 1, 3]), kwargs = {})
%clone_106 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_118,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_179 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_106, [2, 1024, 512]), kwargs = {})
%_param_constant481 : [#users=1] = get_attr[target=_param_constant481]
%t_157 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant481,), kwargs = {})
%view_411 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_179, [2048, 512]), kwargs = {})
%_param_constant482 : [#users=1] = get_attr[target=_param_constant482]
%addmm_73 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant482, %view_411, %t_157), kwargs = {})
%view_412 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_73, [2, 1024, 512]), kwargs = {})
%add_242 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_412, %add_239), kwargs = {})
%convert_element_type_238 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_242, torch.float32), kwargs = {})
%var_mean_77 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_238, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_154 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_77, 0), kwargs = {})
%getitem_155 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_77, 1), kwargs = {})
%add_243 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_154, 1e-05), kwargs = {})
%rsqrt_77 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_243,), kwargs = {})
%sub_77 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_242, %getitem_155), kwargs = {})
%mul_168 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_77, %rsqrt_77), kwargs = {})
%_param_constant483 : [#users=1] = get_attr[target=_param_constant483]
%mul_169 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_168, %_param_constant483), kwargs = {})
%_param_constant484 : [#users=1] = get_attr[target=_param_constant484]
%add_244 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_169, %_param_constant484), kwargs = {})
%convert_element_type_239 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_244, torch.float16), kwargs = {})
%_param_constant485 : [#users=1] = get_attr[target=_param_constant485]
%t_158 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant485,), kwargs = {})
%view_413 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_239, [2048, 512]), kwargs = {})
%_param_constant486 : [#users=1] = get_attr[target=_param_constant486]
%addmm_74 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant486, %view_413, %t_158), kwargs = {})
%view_414 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%addmm_74, [2, 1024, 4096]), kwargs = {})
%slice_59 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_414, -1, 0, 2048), kwargs = {})
%slice_60 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_414, -1, 2048, 4096), kwargs = {})
%gelu_11 : [#users=1] = call_function[target=torch.ops.aten.gelu](args = (%slice_60,), kwargs = {})
%mul_170 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%slice_59, %gelu_11), kwargs = {})
%_param_constant487 : [#users=1] = get_attr[target=_param_constant487]
%t_159 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant487,), kwargs = {})
%view_415 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_170, [2048, 2048]), kwargs = {})
%_param_constant488 : [#users=1] = get_attr[target=_param_constant488]
%addmm_75 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant488, %view_415, %t_159), kwargs = {})
%view_416 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_75, [2, 1024, 512]), kwargs = {})
%add_245 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%view_416, %add_242), kwargs = {})
%_param_constant489 : [#users=1] = get_attr[target=_param_constant489]
%t_160 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant489,), kwargs = {})
%view_417 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%add_245, [2048, 512]), kwargs = {})
%_param_constant490 : [#users=1] = get_attr[target=_param_constant490]
%addmm_76 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant490, %view_417, %t_160), kwargs = {})
%view_418 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_76, [2, 1024, 512]), kwargs = {})
%view_419 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%view_418, [2, 32, 32, 512]), kwargs = {})
%permute_119 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_419, [0, 3, 1, 2]), kwargs = {})
%clone_107 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_119,), kwargs = {memory_format: torch.contiguous_format})
%add_246 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%clone_107, %div_15), kwargs = {})
%cat_7 : [#users=2] = call_function[target=torch.ops.aten.cat](args = ([%add_246, %convolution_11], 1), kwargs = {})
%view_420 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%cat_7, [2, 32, 32, 1024]), kwargs = {})
%convert_element_type_240 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_420, torch.float32), kwargs = {})
%var_mean_78 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_240, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_156 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_78, 0), kwargs = {})
%getitem_157 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_78, 1), kwargs = {})
%add_247 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_156, 1e-05), kwargs = {})
%rsqrt_78 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_247,), kwargs = {})
%sub_78 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_420, %getitem_157), kwargs = {})
%mul_171 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_78, %rsqrt_78), kwargs = {})
%view_421 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_171, [2, 1024, 32, 32]), kwargs = {})
%_param_constant491 : [#users=1] = get_attr[target=_param_constant491]
%unsqueeze_284 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant491, 0), kwargs = {})
%unsqueeze_285 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_284, 2), kwargs = {})
%unsqueeze_286 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_285, 3), kwargs = {})
%_param_constant492 : [#users=1] = get_attr[target=_param_constant492]
%unsqueeze_287 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant492, 0), kwargs = {})
%unsqueeze_288 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_287, 2), kwargs = {})
%unsqueeze_289 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_288, 3), kwargs = {})
%mul_172 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_421, %unsqueeze_289), kwargs = {})
%add_248 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_172, %unsqueeze_286), kwargs = {})
%convert_element_type_241 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_248, torch.float16), kwargs = {})
%convert_element_type_242 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_157, torch.float16), kwargs = {})
%convert_element_type_243 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_78, torch.float16), kwargs = {})
%squeeze_168 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_242, 3), kwargs = {})
%squeeze_169 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_168, 2), kwargs = {})
%squeeze_170 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_243, 3), kwargs = {})
%squeeze_171 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_170, 2), kwargs = {})
%detach_108 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_169,), kwargs = {})
%detach_109 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_171,), kwargs = {})
%silu_46 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_241,), kwargs = {})
%_param_constant493 : [#users=1] = get_attr[target=_param_constant493]
%_param_constant494 : [#users=1] = get_attr[target=_param_constant494]
%convolution_42 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_46, %_param_constant493, %_param_constant494, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_47 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant495 : [#users=1] = get_attr[target=_param_constant495]
%t_161 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant495,), kwargs = {})
%_param_constant496 : [#users=1] = get_attr[target=_param_constant496]
%addmm_77 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant496, %silu_47, %t_161), kwargs = {})
%slice_61 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_77, 0, 0, 9223372036854775807), kwargs = {})
%slice_62 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_61, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_290 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_62, 2), kwargs = {})
%unsqueeze_291 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_290, 3), kwargs = {})
%add_249 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_42, %unsqueeze_291), kwargs = {})
%view_422 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_249, [2, 32, 16, 1024]), kwargs = {})
%convert_element_type_244 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_422, torch.float32), kwargs = {})
%var_mean_79 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_244, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_158 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_79, 0), kwargs = {})
%getitem_159 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_79, 1), kwargs = {})
%add_250 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_158, 1e-05), kwargs = {})
%rsqrt_79 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_250,), kwargs = {})
%sub_79 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_422, %getitem_159), kwargs = {})
%mul_173 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_79, %rsqrt_79), kwargs = {})
%view_423 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_173, [2, 512, 32, 32]), kwargs = {})
%_param_constant497 : [#users=1] = get_attr[target=_param_constant497]
%unsqueeze_292 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant497, 0), kwargs = {})
%unsqueeze_293 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_292, 2), kwargs = {})
%unsqueeze_294 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_293, 3), kwargs = {})
%_param_constant498 : [#users=1] = get_attr[target=_param_constant498]
%unsqueeze_295 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant498, 0), kwargs = {})
%unsqueeze_296 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_295, 2), kwargs = {})
%unsqueeze_297 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_296, 3), kwargs = {})
%mul_174 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_423, %unsqueeze_297), kwargs = {})
%add_251 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_174, %unsqueeze_294), kwargs = {})
%convert_element_type_245 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_251, torch.float16), kwargs = {})
%convert_element_type_246 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_159, torch.float16), kwargs = {})
%convert_element_type_247 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_79, torch.float16), kwargs = {})
%squeeze_172 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_246, 3), kwargs = {})
%squeeze_173 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_172, 2), kwargs = {})
%squeeze_174 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_247, 3), kwargs = {})
%squeeze_175 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_174, 2), kwargs = {})
%detach_110 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_173,), kwargs = {})
%detach_111 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_175,), kwargs = {})
%silu_48 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_245,), kwargs = {})
%_param_constant499 : [#users=1] = get_attr[target=_param_constant499]
%_param_constant500 : [#users=1] = get_attr[target=_param_constant500]
%convolution_43 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_48, %_param_constant499, %_param_constant500, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%_param_constant501 : [#users=1] = get_attr[target=_param_constant501]
%_param_constant502 : [#users=1] = get_attr[target=_param_constant502]
%convolution_44 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%cat_7, %_param_constant501, %_param_constant502, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
%add_252 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_44, %convolution_43), kwargs = {})
%div_16 : [#users=2] = call_function[target=torch.ops.aten.div](args = (%add_252, 1.0), kwargs = {})
%view_424 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%div_16, [2, 32, 16, 1024]), kwargs = {})
%convert_element_type_248 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_424, torch.float32), kwargs = {})
%var_mean_80 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_248, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_160 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_80, 0), kwargs = {})
%getitem_161 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_80, 1), kwargs = {})
%add_253 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_160, 1e-06), kwargs = {})
%rsqrt_80 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_253,), kwargs = {})
%sub_80 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_424, %getitem_161), kwargs = {})
%mul_175 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_80, %rsqrt_80), kwargs = {})
%view_425 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_175, [2, 512, 32, 32]), kwargs = {})
%_param_constant503 : [#users=1] = get_attr[target=_param_constant503]
%unsqueeze_298 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant503, 0), kwargs = {})
%unsqueeze_299 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_298, 2), kwargs = {})
%unsqueeze_300 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_299, 3), kwargs = {})
%_param_constant504 : [#users=1] = get_attr[target=_param_constant504]
%unsqueeze_301 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant504, 0), kwargs = {})
%unsqueeze_302 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_301, 2), kwargs = {})
%unsqueeze_303 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_302, 3), kwargs = {})
%mul_176 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_425, %unsqueeze_303), kwargs = {})
%add_254 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_176, %unsqueeze_300), kwargs = {})
%convert_element_type_249 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_254, torch.float16), kwargs = {})
%convert_element_type_250 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_161, torch.float16), kwargs = {})
%convert_element_type_251 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_80, torch.float16), kwargs = {})
%squeeze_176 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_250, 3), kwargs = {})
%squeeze_177 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_176, 2), kwargs = {})
%squeeze_178 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_251, 3), kwargs = {})
%squeeze_179 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_178, 2), kwargs = {})
%detach_112 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_177,), kwargs = {})
%detach_113 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_179,), kwargs = {})
%permute_120 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%convert_element_type_249, [0, 2, 3, 1]), kwargs = {})
%view_426 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%permute_120, [2, 1024, 512]), kwargs = {})
%_param_constant505 : [#users=1] = get_attr[target=_param_constant505]
%t_162 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant505,), kwargs = {})
%expand_25 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%view_426, [2, 1024, 512]), kwargs = {})
%view_427 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_25, [2, 1024, 512]), kwargs = {})
%expand_26 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%t_162, [2, 512, 512]), kwargs = {})
%view_428 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_26, [2, 512, 512]), kwargs = {})
%bmm_36 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%view_427, %view_428), kwargs = {})
%_unsafe_view_180 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%bmm_36, [2, 1024, 512]), kwargs = {})
%_param_constant506 : [#users=1] = get_attr[target=_param_constant506]
%add_255 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%_unsafe_view_180, %_param_constant506), kwargs = {})
%convert_element_type_252 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_255, torch.float32), kwargs = {})
%var_mean_81 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_252, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_162 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_81, 0), kwargs = {})
%getitem_163 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_81, 1), kwargs = {})
%add_256 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_162, 1e-05), kwargs = {})
%rsqrt_81 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_256,), kwargs = {})
%sub_81 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_255, %getitem_163), kwargs = {})
%mul_177 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_81, %rsqrt_81), kwargs = {})
%_param_constant507 : [#users=1] = get_attr[target=_param_constant507]
%mul_178 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_177, %_param_constant507), kwargs = {})
%_param_constant508 : [#users=1] = get_attr[target=_param_constant508]
%add_257 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_178, %_param_constant508), kwargs = {})
%convert_element_type_253 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_257, torch.float16), kwargs = {})
%_param_constant509 : [#users=1] = get_attr[target=_param_constant509]
%t_163 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant509,), kwargs = {})
%view_429 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_253, [2048, 512]), kwargs = {})
%mm_72 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_429, %t_163), kwargs = {})
%_unsafe_view_181 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_72, [2, 1024, 512]), kwargs = {})
%view_430 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_181, [2, 1024, 8, 64]), kwargs = {})
%permute_121 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_430, [0, 2, 1, 3]), kwargs = {})
%clone_108 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_121,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_182 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_108, [16, 1024, 64]), kwargs = {})
%_param_constant510 : [#users=1] = get_attr[target=_param_constant510]
%t_164 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant510,), kwargs = {})
%view_431 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_73 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_431, %t_164), kwargs = {})
%_unsafe_view_183 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_73, [2, 77, 512]), kwargs = {})
%_param_constant511 : [#users=1] = get_attr[target=_param_constant511]
%t_165 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant511,), kwargs = {})
%view_432 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_74 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_432, %t_165), kwargs = {})
%_unsafe_view_184 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_74, [2, 77, 512]), kwargs = {})
%view_433 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_183, [2, 77, 8, 64]), kwargs = {})
%permute_122 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_433, [0, 2, 1, 3]), kwargs = {})
%clone_109 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_122,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_185 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_109, [16, 77, 64]), kwargs = {})
%view_434 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_184, [2, 77, 8, 64]), kwargs = {})
%permute_123 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_434, [0, 2, 1, 3]), kwargs = {})
%clone_110 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_123,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_186 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_110, [16, 77, 64]), kwargs = {})
%empty_24 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 1024, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_24 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_185, -1, -2), kwargs = {})
%baddbmm_24 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_24, %_unsafe_view_182, %transpose_24), kwargs = {beta: 0, alpha: 0.125})
%_softmax_24 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_24, -1, False), kwargs = {})
%detach_114 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_24,), kwargs = {})
%bmm_37 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_24, %_unsafe_view_186), kwargs = {})
%view_435 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_37, [2, 8, 1024, 64]), kwargs = {})
%permute_124 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_435, [0, 2, 1, 3]), kwargs = {})
%clone_111 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_124,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_187 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_111, [2, 1024, 512]), kwargs = {})
%_param_constant512 : [#users=1] = get_attr[target=_param_constant512]
%t_166 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant512,), kwargs = {})
%view_436 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_187, [2048, 512]), kwargs = {})
%_param_constant513 : [#users=1] = get_attr[target=_param_constant513]
%addmm_78 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant513, %view_436, %t_166), kwargs = {})
%view_437 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_78, [2, 1024, 512]), kwargs = {})
%add_258 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_437, %add_255), kwargs = {})
%convert_element_type_254 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_258, torch.float32), kwargs = {})
%var_mean_82 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_254, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_164 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_82, 0), kwargs = {})
%getitem_165 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_82, 1), kwargs = {})
%add_259 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_164, 1e-05), kwargs = {})
%rsqrt_82 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_259,), kwargs = {})
%sub_82 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_258, %getitem_165), kwargs = {})
%mul_179 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_82, %rsqrt_82), kwargs = {})
%_param_constant514 : [#users=1] = get_attr[target=_param_constant514]
%mul_180 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_179, %_param_constant514), kwargs = {})
%_param_constant515 : [#users=1] = get_attr[target=_param_constant515]
%add_260 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_180, %_param_constant515), kwargs = {})
%convert_element_type_255 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_260, torch.float16), kwargs = {})
%_param_constant516 : [#users=1] = get_attr[target=_param_constant516]
%t_167 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant516,), kwargs = {})
%view_438 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_255, [2048, 512]), kwargs = {})
%mm_75 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_438, %t_167), kwargs = {})
%_unsafe_view_188 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_75, [2, 1024, 512]), kwargs = {})
%view_439 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_188, [2, 1024, 8, 64]), kwargs = {})
%permute_125 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_439, [0, 2, 1, 3]), kwargs = {})
%clone_112 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_125,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_189 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_112, [16, 1024, 64]), kwargs = {})
%_param_constant517 : [#users=1] = get_attr[target=_param_constant517]
%t_168 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant517,), kwargs = {})
%view_440 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_76 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_440, %t_168), kwargs = {})
%_unsafe_view_190 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_76, [2, 77, 512]), kwargs = {})
%_param_constant518 : [#users=1] = get_attr[target=_param_constant518]
%t_169 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant518,), kwargs = {})
%view_441 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_77 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_441, %t_169), kwargs = {})
%_unsafe_view_191 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_77, [2, 77, 512]), kwargs = {})
%view_442 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_190, [2, 77, 8, 64]), kwargs = {})
%permute_126 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_442, [0, 2, 1, 3]), kwargs = {})
%clone_113 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_126,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_192 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_113, [16, 77, 64]), kwargs = {})
%view_443 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_191, [2, 77, 8, 64]), kwargs = {})
%permute_127 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_443, [0, 2, 1, 3]), kwargs = {})
%clone_114 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_127,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_193 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_114, [16, 77, 64]), kwargs = {})
%empty_25 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 1024, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_25 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_192, -1, -2), kwargs = {})
%baddbmm_25 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_25, %_unsafe_view_189, %transpose_25), kwargs = {beta: 0, alpha: 0.125})
%_softmax_25 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_25, -1, False), kwargs = {})
%detach_115 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_25,), kwargs = {})
%bmm_38 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_25, %_unsafe_view_193), kwargs = {})
%view_444 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_38, [2, 8, 1024, 64]), kwargs = {})
%permute_128 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_444, [0, 2, 1, 3]), kwargs = {})
%clone_115 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_128,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_194 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_115, [2, 1024, 512]), kwargs = {})
%_param_constant519 : [#users=1] = get_attr[target=_param_constant519]
%t_170 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant519,), kwargs = {})
%view_445 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_194, [2048, 512]), kwargs = {})
%_param_constant520 : [#users=1] = get_attr[target=_param_constant520]
%addmm_79 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant520, %view_445, %t_170), kwargs = {})
%view_446 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_79, [2, 1024, 512]), kwargs = {})
%add_261 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_446, %add_258), kwargs = {})
%convert_element_type_256 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_261, torch.float32), kwargs = {})
%var_mean_83 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_256, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_166 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_83, 0), kwargs = {})
%getitem_167 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_83, 1), kwargs = {})
%add_262 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_166, 1e-05), kwargs = {})
%rsqrt_83 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_262,), kwargs = {})
%sub_83 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_261, %getitem_167), kwargs = {})
%mul_181 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_83, %rsqrt_83), kwargs = {})
%_param_constant521 : [#users=1] = get_attr[target=_param_constant521]
%mul_182 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_181, %_param_constant521), kwargs = {})
%_param_constant522 : [#users=1] = get_attr[target=_param_constant522]
%add_263 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_182, %_param_constant522), kwargs = {})
%convert_element_type_257 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_263, torch.float16), kwargs = {})
%_param_constant523 : [#users=1] = get_attr[target=_param_constant523]
%t_171 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant523,), kwargs = {})
%view_447 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_257, [2048, 512]), kwargs = {})
%_param_constant524 : [#users=1] = get_attr[target=_param_constant524]
%addmm_80 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant524, %view_447, %t_171), kwargs = {})
%view_448 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%addmm_80, [2, 1024, 4096]), kwargs = {})
%slice_63 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_448, -1, 0, 2048), kwargs = {})
%slice_64 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_448, -1, 2048, 4096), kwargs = {})
%gelu_12 : [#users=1] = call_function[target=torch.ops.aten.gelu](args = (%slice_64,), kwargs = {})
%mul_183 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%slice_63, %gelu_12), kwargs = {})
%_param_constant525 : [#users=1] = get_attr[target=_param_constant525]
%t_172 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant525,), kwargs = {})
%view_449 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_183, [2048, 2048]), kwargs = {})
%_param_constant526 : [#users=1] = get_attr[target=_param_constant526]
%addmm_81 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant526, %view_449, %t_172), kwargs = {})
%view_450 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_81, [2, 1024, 512]), kwargs = {})
%add_264 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%view_450, %add_261), kwargs = {})
%_param_constant527 : [#users=1] = get_attr[target=_param_constant527]
%t_173 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant527,), kwargs = {})
%view_451 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%add_264, [2048, 512]), kwargs = {})
%_param_constant528 : [#users=1] = get_attr[target=_param_constant528]
%addmm_82 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant528, %view_451, %t_173), kwargs = {})
%view_452 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_82, [2, 1024, 512]), kwargs = {})
%view_453 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%view_452, [2, 32, 32, 512]), kwargs = {})
%permute_129 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_453, [0, 3, 1, 2]), kwargs = {})
%clone_116 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_129,), kwargs = {memory_format: torch.contiguous_format})
%add_265 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%clone_116, %div_16), kwargs = {})
%upsample_nearest2d_1 : [#users=1] = call_function[target=torch.ops.aten.upsample_nearest2d](args = (%add_265, [64, 64], 2.0, 2.0), kwargs = {})
%_param_constant529 : [#users=1] = get_attr[target=_param_constant529]
%_param_constant530 : [#users=1] = get_attr[target=_param_constant530]
%convolution_45 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%upsample_nearest2d_1, %_param_constant529, %_param_constant530, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%cat_8 : [#users=2] = call_function[target=torch.ops.aten.cat](args = ([%convolution_45, %add_50], 1), kwargs = {})
%view_454 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%cat_8, [2, 32, 32, 4096]), kwargs = {})
%convert_element_type_258 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_454, torch.float32), kwargs = {})
%var_mean_84 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_258, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_168 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_84, 0), kwargs = {})
%getitem_169 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_84, 1), kwargs = {})
%add_266 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_168, 1e-05), kwargs = {})
%rsqrt_84 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_266,), kwargs = {})
%sub_84 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_454, %getitem_169), kwargs = {})
%mul_184 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_84, %rsqrt_84), kwargs = {})
%view_455 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_184, [2, 1024, 64, 64]), kwargs = {})
%_param_constant531 : [#users=1] = get_attr[target=_param_constant531]
%unsqueeze_304 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant531, 0), kwargs = {})
%unsqueeze_305 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_304, 2), kwargs = {})
%unsqueeze_306 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_305, 3), kwargs = {})
%_param_constant532 : [#users=1] = get_attr[target=_param_constant532]
%unsqueeze_307 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant532, 0), kwargs = {})
%unsqueeze_308 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_307, 2), kwargs = {})
%unsqueeze_309 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_308, 3), kwargs = {})
%mul_185 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_455, %unsqueeze_309), kwargs = {})
%add_267 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_185, %unsqueeze_306), kwargs = {})
%convert_element_type_259 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_267, torch.float16), kwargs = {})
%convert_element_type_260 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_169, torch.float16), kwargs = {})
%convert_element_type_261 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_84, torch.float16), kwargs = {})
%squeeze_180 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_260, 3), kwargs = {})
%squeeze_181 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_180, 2), kwargs = {})
%squeeze_182 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_261, 3), kwargs = {})
%squeeze_183 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_182, 2), kwargs = {})
%detach_116 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_181,), kwargs = {})
%detach_117 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_183,), kwargs = {})
%silu_49 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_259,), kwargs = {})
%_param_constant533 : [#users=1] = get_attr[target=_param_constant533]
%_param_constant534 : [#users=1] = get_attr[target=_param_constant534]
%convolution_46 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_49, %_param_constant533, %_param_constant534, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_50 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant535 : [#users=1] = get_attr[target=_param_constant535]
%t_174 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant535,), kwargs = {})
%_param_constant536 : [#users=1] = get_attr[target=_param_constant536]
%addmm_83 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant536, %silu_50, %t_174), kwargs = {})
%slice_65 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_83, 0, 0, 9223372036854775807), kwargs = {})
%slice_66 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_65, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_310 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_66, 2), kwargs = {})
%unsqueeze_311 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_310, 3), kwargs = {})
%add_268 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_46, %unsqueeze_311), kwargs = {})
%view_456 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_268, [2, 32, 16, 4096]), kwargs = {})
%convert_element_type_262 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_456, torch.float32), kwargs = {})
%var_mean_85 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_262, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_170 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_85, 0), kwargs = {})
%getitem_171 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_85, 1), kwargs = {})
%add_269 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_170, 1e-05), kwargs = {})
%rsqrt_85 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_269,), kwargs = {})
%sub_85 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_456, %getitem_171), kwargs = {})
%mul_186 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_85, %rsqrt_85), kwargs = {})
%view_457 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_186, [2, 512, 64, 64]), kwargs = {})
%_param_constant537 : [#users=1] = get_attr[target=_param_constant537]
%unsqueeze_312 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant537, 0), kwargs = {})
%unsqueeze_313 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_312, 2), kwargs = {})
%unsqueeze_314 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_313, 3), kwargs = {})
%_param_constant538 : [#users=1] = get_attr[target=_param_constant538]
%unsqueeze_315 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant538, 0), kwargs = {})
%unsqueeze_316 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_315, 2), kwargs = {})
%unsqueeze_317 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_316, 3), kwargs = {})
%mul_187 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_457, %unsqueeze_317), kwargs = {})
%add_270 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_187, %unsqueeze_314), kwargs = {})
%convert_element_type_263 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_270, torch.float16), kwargs = {})
%convert_element_type_264 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_171, torch.float16), kwargs = {})
%convert_element_type_265 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_85, torch.float16), kwargs = {})
%squeeze_184 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_264, 3), kwargs = {})
%squeeze_185 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_184, 2), kwargs = {})
%squeeze_186 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_265, 3), kwargs = {})
%squeeze_187 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_186, 2), kwargs = {})
%detach_118 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_185,), kwargs = {})
%detach_119 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_187,), kwargs = {})
%silu_51 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_263,), kwargs = {})
%_param_constant539 : [#users=1] = get_attr[target=_param_constant539]
%_param_constant540 : [#users=1] = get_attr[target=_param_constant540]
%convolution_47 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_51, %_param_constant539, %_param_constant540, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%_param_constant541 : [#users=1] = get_attr[target=_param_constant541]
%_param_constant542 : [#users=1] = get_attr[target=_param_constant542]
%convolution_48 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%cat_8, %_param_constant541, %_param_constant542, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
%add_271 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_48, %convolution_47), kwargs = {})
%div_17 : [#users=2] = call_function[target=torch.ops.aten.div](args = (%add_271, 1.0), kwargs = {})
%view_458 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%div_17, [2, 32, 16, 4096]), kwargs = {})
%convert_element_type_266 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_458, torch.float32), kwargs = {})
%var_mean_86 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_266, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_172 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_86, 0), kwargs = {})
%getitem_173 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_86, 1), kwargs = {})
%add_272 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_172, 1e-06), kwargs = {})
%rsqrt_86 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_272,), kwargs = {})
%sub_86 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_458, %getitem_173), kwargs = {})
%mul_188 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_86, %rsqrt_86), kwargs = {})
%view_459 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_188, [2, 512, 64, 64]), kwargs = {})
%_param_constant543 : [#users=1] = get_attr[target=_param_constant543]
%unsqueeze_318 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant543, 0), kwargs = {})
%unsqueeze_319 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_318, 2), kwargs = {})
%unsqueeze_320 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_319, 3), kwargs = {})
%_param_constant544 : [#users=1] = get_attr[target=_param_constant544]
%unsqueeze_321 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant544, 0), kwargs = {})
%unsqueeze_322 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_321, 2), kwargs = {})
%unsqueeze_323 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_322, 3), kwargs = {})
%mul_189 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_459, %unsqueeze_323), kwargs = {})
%add_273 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_189, %unsqueeze_320), kwargs = {})
%convert_element_type_267 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_273, torch.float16), kwargs = {})
%convert_element_type_268 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_173, torch.float16), kwargs = {})
%convert_element_type_269 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_86, torch.float16), kwargs = {})
%squeeze_188 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_268, 3), kwargs = {})
%squeeze_189 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_188, 2), kwargs = {})
%squeeze_190 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_269, 3), kwargs = {})
%squeeze_191 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_190, 2), kwargs = {})
%detach_120 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_189,), kwargs = {})
%detach_121 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_191,), kwargs = {})
%permute_130 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%convert_element_type_267, [0, 2, 3, 1]), kwargs = {})
%view_460 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%permute_130, [2, 4096, 512]), kwargs = {})
%_param_constant545 : [#users=1] = get_attr[target=_param_constant545]
%t_175 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant545,), kwargs = {})
%expand_27 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%view_460, [2, 4096, 512]), kwargs = {})
%view_461 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_27, [2, 4096, 512]), kwargs = {})
%expand_28 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%t_175, [2, 512, 512]), kwargs = {})
%view_462 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_28, [2, 512, 512]), kwargs = {})
%bmm_39 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%view_461, %view_462), kwargs = {})
%_unsafe_view_195 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%bmm_39, [2, 4096, 512]), kwargs = {})
%_param_constant546 : [#users=1] = get_attr[target=_param_constant546]
%add_274 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%_unsafe_view_195, %_param_constant546), kwargs = {})
%convert_element_type_270 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_274, torch.float32), kwargs = {})
%var_mean_87 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_270, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_174 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_87, 0), kwargs = {})
%getitem_175 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_87, 1), kwargs = {})
%add_275 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_174, 1e-05), kwargs = {})
%rsqrt_87 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_275,), kwargs = {})
%sub_87 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_274, %getitem_175), kwargs = {})
%mul_190 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_87, %rsqrt_87), kwargs = {})
%_param_constant547 : [#users=1] = get_attr[target=_param_constant547]
%mul_191 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_190, %_param_constant547), kwargs = {})
%_param_constant548 : [#users=1] = get_attr[target=_param_constant548]
%add_276 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_191, %_param_constant548), kwargs = {})
%convert_element_type_271 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_276, torch.float16), kwargs = {})
%_param_constant549 : [#users=1] = get_attr[target=_param_constant549]
%t_176 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant549,), kwargs = {})
%view_463 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_271, [8192, 512]), kwargs = {})
%mm_78 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_463, %t_176), kwargs = {})
%_unsafe_view_196 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_78, [2, 4096, 512]), kwargs = {})
%view_464 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_196, [2, 4096, 8, 64]), kwargs = {})
%permute_131 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_464, [0, 2, 1, 3]), kwargs = {})
%clone_117 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_131,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_197 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_117, [16, 4096, 64]), kwargs = {})
%_param_constant550 : [#users=1] = get_attr[target=_param_constant550]
%t_177 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant550,), kwargs = {})
%view_465 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_79 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_465, %t_177), kwargs = {})
%_unsafe_view_198 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_79, [2, 77, 512]), kwargs = {})
%_param_constant551 : [#users=1] = get_attr[target=_param_constant551]
%t_178 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant551,), kwargs = {})
%view_466 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_80 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_466, %t_178), kwargs = {})
%_unsafe_view_199 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_80, [2, 77, 512]), kwargs = {})
%view_467 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_198, [2, 77, 8, 64]), kwargs = {})
%permute_132 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_467, [0, 2, 1, 3]), kwargs = {})
%clone_118 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_132,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_200 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_118, [16, 77, 64]), kwargs = {})
%view_468 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_199, [2, 77, 8, 64]), kwargs = {})
%permute_133 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_468, [0, 2, 1, 3]), kwargs = {})
%clone_119 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_133,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_201 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_119, [16, 77, 64]), kwargs = {})
%empty_26 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 4096, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_26 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_200, -1, -2), kwargs = {})
%baddbmm_26 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_26, %_unsafe_view_197, %transpose_26), kwargs = {beta: 0, alpha: 0.125})
%_softmax_26 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_26, -1, False), kwargs = {})
%detach_122 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_26,), kwargs = {})
%bmm_40 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_26, %_unsafe_view_201), kwargs = {})
%view_469 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_40, [2, 8, 4096, 64]), kwargs = {})
%permute_134 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_469, [0, 2, 1, 3]), kwargs = {})
%clone_120 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_134,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_202 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_120, [2, 4096, 512]), kwargs = {})
%_param_constant552 : [#users=1] = get_attr[target=_param_constant552]
%t_179 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant552,), kwargs = {})
%view_470 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_202, [8192, 512]), kwargs = {})
%_param_constant553 : [#users=1] = get_attr[target=_param_constant553]
%addmm_84 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant553, %view_470, %t_179), kwargs = {})
%view_471 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_84, [2, 4096, 512]), kwargs = {})
%add_277 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_471, %add_274), kwargs = {})
%convert_element_type_272 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_277, torch.float32), kwargs = {})
%var_mean_88 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_272, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_176 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_88, 0), kwargs = {})
%getitem_177 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_88, 1), kwargs = {})
%add_278 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_176, 1e-05), kwargs = {})
%rsqrt_88 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_278,), kwargs = {})
%sub_88 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_277, %getitem_177), kwargs = {})
%mul_192 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_88, %rsqrt_88), kwargs = {})
%_param_constant554 : [#users=1] = get_attr[target=_param_constant554]
%mul_193 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_192, %_param_constant554), kwargs = {})
%_param_constant555 : [#users=1] = get_attr[target=_param_constant555]
%add_279 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_193, %_param_constant555), kwargs = {})
%convert_element_type_273 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_279, torch.float16), kwargs = {})
%_param_constant556 : [#users=1] = get_attr[target=_param_constant556]
%t_180 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant556,), kwargs = {})
%view_472 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_273, [8192, 512]), kwargs = {})
%mm_81 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_472, %t_180), kwargs = {})
%_unsafe_view_203 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_81, [2, 4096, 512]), kwargs = {})
%view_473 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_203, [2, 4096, 8, 64]), kwargs = {})
%permute_135 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_473, [0, 2, 1, 3]), kwargs = {})
%clone_121 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_135,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_204 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_121, [16, 4096, 64]), kwargs = {})
%_param_constant557 : [#users=1] = get_attr[target=_param_constant557]
%t_181 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant557,), kwargs = {})
%view_474 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_82 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_474, %t_181), kwargs = {})
%_unsafe_view_205 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_82, [2, 77, 512]), kwargs = {})
%_param_constant558 : [#users=1] = get_attr[target=_param_constant558]
%t_182 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant558,), kwargs = {})
%view_475 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_83 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_475, %t_182), kwargs = {})
%_unsafe_view_206 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_83, [2, 77, 512]), kwargs = {})
%view_476 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_205, [2, 77, 8, 64]), kwargs = {})
%permute_136 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_476, [0, 2, 1, 3]), kwargs = {})
%clone_122 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_136,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_207 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_122, [16, 77, 64]), kwargs = {})
%view_477 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_206, [2, 77, 8, 64]), kwargs = {})
%permute_137 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_477, [0, 2, 1, 3]), kwargs = {})
%clone_123 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_137,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_208 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_123, [16, 77, 64]), kwargs = {})
%empty_27 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 4096, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_27 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_207, -1, -2), kwargs = {})
%baddbmm_27 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_27, %_unsafe_view_204, %transpose_27), kwargs = {beta: 0, alpha: 0.125})
%_softmax_27 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_27, -1, False), kwargs = {})
%detach_123 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_27,), kwargs = {})
%bmm_41 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_27, %_unsafe_view_208), kwargs = {})
%view_478 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_41, [2, 8, 4096, 64]), kwargs = {})
%permute_138 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_478, [0, 2, 1, 3]), kwargs = {})
%clone_124 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_138,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_209 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_124, [2, 4096, 512]), kwargs = {})
%_param_constant559 : [#users=1] = get_attr[target=_param_constant559]
%t_183 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant559,), kwargs = {})
%view_479 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_209, [8192, 512]), kwargs = {})
%_param_constant560 : [#users=1] = get_attr[target=_param_constant560]
%addmm_85 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant560, %view_479, %t_183), kwargs = {})
%view_480 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_85, [2, 4096, 512]), kwargs = {})
%add_280 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_480, %add_277), kwargs = {})
%convert_element_type_274 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_280, torch.float32), kwargs = {})
%var_mean_89 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_274, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_178 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_89, 0), kwargs = {})
%getitem_179 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_89, 1), kwargs = {})
%add_281 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_178, 1e-05), kwargs = {})
%rsqrt_89 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_281,), kwargs = {})
%sub_89 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_280, %getitem_179), kwargs = {})
%mul_194 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_89, %rsqrt_89), kwargs = {})
%_param_constant561 : [#users=1] = get_attr[target=_param_constant561]
%mul_195 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_194, %_param_constant561), kwargs = {})
%_param_constant562 : [#users=1] = get_attr[target=_param_constant562]
%add_282 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_195, %_param_constant562), kwargs = {})
%convert_element_type_275 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_282, torch.float16), kwargs = {})
%_param_constant563 : [#users=1] = get_attr[target=_param_constant563]
%t_184 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant563,), kwargs = {})
%view_481 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_275, [8192, 512]), kwargs = {})
%_param_constant564 : [#users=1] = get_attr[target=_param_constant564]
%addmm_86 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant564, %view_481, %t_184), kwargs = {})
%view_482 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%addmm_86, [2, 4096, 4096]), kwargs = {})
%slice_67 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_482, -1, 0, 2048), kwargs = {})
%slice_68 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_482, -1, 2048, 4096), kwargs = {})
%gelu_13 : [#users=1] = call_function[target=torch.ops.aten.gelu](args = (%slice_68,), kwargs = {})
%mul_196 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%slice_67, %gelu_13), kwargs = {})
%_param_constant565 : [#users=1] = get_attr[target=_param_constant565]
%t_185 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant565,), kwargs = {})
%view_483 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_196, [8192, 2048]), kwargs = {})
%_param_constant566 : [#users=1] = get_attr[target=_param_constant566]
%addmm_87 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant566, %view_483, %t_185), kwargs = {})
%view_484 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_87, [2, 4096, 512]), kwargs = {})
%add_283 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%view_484, %add_280), kwargs = {})
%_param_constant567 : [#users=1] = get_attr[target=_param_constant567]
%t_186 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant567,), kwargs = {})
%view_485 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%add_283, [8192, 512]), kwargs = {})
%_param_constant568 : [#users=1] = get_attr[target=_param_constant568]
%addmm_88 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant568, %view_485, %t_186), kwargs = {})
%view_486 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_88, [2, 4096, 512]), kwargs = {})
%view_487 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%view_486, [2, 64, 64, 512]), kwargs = {})
%permute_139 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_487, [0, 3, 1, 2]), kwargs = {})
%clone_125 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_139,), kwargs = {memory_format: torch.contiguous_format})
%add_284 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%clone_125, %div_17), kwargs = {})
%cat_9 : [#users=2] = call_function[target=torch.ops.aten.cat](args = ([%add_284, %add_31], 1), kwargs = {})
%view_488 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%cat_9, [2, 32, 32, 4096]), kwargs = {})
%convert_element_type_276 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_488, torch.float32), kwargs = {})
%var_mean_90 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_276, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_180 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_90, 0), kwargs = {})
%getitem_181 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_90, 1), kwargs = {})
%add_285 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_180, 1e-05), kwargs = {})
%rsqrt_90 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_285,), kwargs = {})
%sub_90 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_488, %getitem_181), kwargs = {})
%mul_197 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_90, %rsqrt_90), kwargs = {})
%view_489 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_197, [2, 1024, 64, 64]), kwargs = {})
%_param_constant569 : [#users=1] = get_attr[target=_param_constant569]
%unsqueeze_324 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant569, 0), kwargs = {})
%unsqueeze_325 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_324, 2), kwargs = {})
%unsqueeze_326 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_325, 3), kwargs = {})
%_param_constant570 : [#users=1] = get_attr[target=_param_constant570]
%unsqueeze_327 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant570, 0), kwargs = {})
%unsqueeze_328 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_327, 2), kwargs = {})
%unsqueeze_329 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_328, 3), kwargs = {})
%mul_198 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_489, %unsqueeze_329), kwargs = {})
%add_286 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_198, %unsqueeze_326), kwargs = {})
%convert_element_type_277 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_286, torch.float16), kwargs = {})
%convert_element_type_278 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_181, torch.float16), kwargs = {})
%convert_element_type_279 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_90, torch.float16), kwargs = {})
%squeeze_192 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_278, 3), kwargs = {})
%squeeze_193 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_192, 2), kwargs = {})
%squeeze_194 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_279, 3), kwargs = {})
%squeeze_195 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_194, 2), kwargs = {})
%detach_124 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_193,), kwargs = {})
%detach_125 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_195,), kwargs = {})
%silu_52 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_277,), kwargs = {})
%_param_constant571 : [#users=1] = get_attr[target=_param_constant571]
%_param_constant572 : [#users=1] = get_attr[target=_param_constant572]
%convolution_49 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_52, %_param_constant571, %_param_constant572, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_53 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant573 : [#users=1] = get_attr[target=_param_constant573]
%t_187 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant573,), kwargs = {})
%_param_constant574 : [#users=1] = get_attr[target=_param_constant574]
%addmm_89 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant574, %silu_53, %t_187), kwargs = {})
%slice_69 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_89, 0, 0, 9223372036854775807), kwargs = {})
%slice_70 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_69, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_330 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_70, 2), kwargs = {})
%unsqueeze_331 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_330, 3), kwargs = {})
%add_287 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_49, %unsqueeze_331), kwargs = {})
%view_490 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_287, [2, 32, 16, 4096]), kwargs = {})
%convert_element_type_280 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_490, torch.float32), kwargs = {})
%var_mean_91 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_280, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_182 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_91, 0), kwargs = {})
%getitem_183 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_91, 1), kwargs = {})
%add_288 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_182, 1e-05), kwargs = {})
%rsqrt_91 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_288,), kwargs = {})
%sub_91 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_490, %getitem_183), kwargs = {})
%mul_199 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_91, %rsqrt_91), kwargs = {})
%view_491 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_199, [2, 512, 64, 64]), kwargs = {})
%_param_constant575 : [#users=1] = get_attr[target=_param_constant575]
%unsqueeze_332 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant575, 0), kwargs = {})
%unsqueeze_333 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_332, 2), kwargs = {})
%unsqueeze_334 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_333, 3), kwargs = {})
%_param_constant576 : [#users=1] = get_attr[target=_param_constant576]
%unsqueeze_335 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant576, 0), kwargs = {})
%unsqueeze_336 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_335, 2), kwargs = {})
%unsqueeze_337 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_336, 3), kwargs = {})
%mul_200 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_491, %unsqueeze_337), kwargs = {})
%add_289 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_200, %unsqueeze_334), kwargs = {})
%convert_element_type_281 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_289, torch.float16), kwargs = {})
%convert_element_type_282 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_183, torch.float16), kwargs = {})
%convert_element_type_283 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_91, torch.float16), kwargs = {})
%squeeze_196 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_282, 3), kwargs = {})
%squeeze_197 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_196, 2), kwargs = {})
%squeeze_198 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_283, 3), kwargs = {})
%squeeze_199 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_198, 2), kwargs = {})
%detach_126 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_197,), kwargs = {})
%detach_127 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_199,), kwargs = {})
%silu_54 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_281,), kwargs = {})
%_param_constant577 : [#users=1] = get_attr[target=_param_constant577]
%_param_constant578 : [#users=1] = get_attr[target=_param_constant578]
%convolution_50 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_54, %_param_constant577, %_param_constant578, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%_param_constant579 : [#users=1] = get_attr[target=_param_constant579]
%_param_constant580 : [#users=1] = get_attr[target=_param_constant580]
%convolution_51 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%cat_9, %_param_constant579, %_param_constant580, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
%add_290 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_51, %convolution_50), kwargs = {})
%div_18 : [#users=2] = call_function[target=torch.ops.aten.div](args = (%add_290, 1.0), kwargs = {})
%view_492 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%div_18, [2, 32, 16, 4096]), kwargs = {})
%convert_element_type_284 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_492, torch.float32), kwargs = {})
%var_mean_92 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_284, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_184 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_92, 0), kwargs = {})
%getitem_185 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_92, 1), kwargs = {})
%add_291 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_184, 1e-06), kwargs = {})
%rsqrt_92 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_291,), kwargs = {})
%sub_92 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_492, %getitem_185), kwargs = {})
%mul_201 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_92, %rsqrt_92), kwargs = {})
%view_493 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_201, [2, 512, 64, 64]), kwargs = {})
%_param_constant581 : [#users=1] = get_attr[target=_param_constant581]
%unsqueeze_338 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant581, 0), kwargs = {})
%unsqueeze_339 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_338, 2), kwargs = {})
%unsqueeze_340 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_339, 3), kwargs = {})
%_param_constant582 : [#users=1] = get_attr[target=_param_constant582]
%unsqueeze_341 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant582, 0), kwargs = {})
%unsqueeze_342 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_341, 2), kwargs = {})
%unsqueeze_343 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_342, 3), kwargs = {})
%mul_202 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_493, %unsqueeze_343), kwargs = {})
%add_292 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_202, %unsqueeze_340), kwargs = {})
%convert_element_type_285 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_292, torch.float16), kwargs = {})
%convert_element_type_286 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_185, torch.float16), kwargs = {})
%convert_element_type_287 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_92, torch.float16), kwargs = {})
%squeeze_200 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_286, 3), kwargs = {})
%squeeze_201 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_200, 2), kwargs = {})
%squeeze_202 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_287, 3), kwargs = {})
%squeeze_203 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_202, 2), kwargs = {})
%detach_128 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_201,), kwargs = {})
%detach_129 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_203,), kwargs = {})
%permute_140 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%convert_element_type_285, [0, 2, 3, 1]), kwargs = {})
%view_494 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%permute_140, [2, 4096, 512]), kwargs = {})
%_param_constant583 : [#users=1] = get_attr[target=_param_constant583]
%t_188 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant583,), kwargs = {})
%expand_29 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%view_494, [2, 4096, 512]), kwargs = {})
%view_495 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_29, [2, 4096, 512]), kwargs = {})
%expand_30 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%t_188, [2, 512, 512]), kwargs = {})
%view_496 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_30, [2, 512, 512]), kwargs = {})
%bmm_42 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%view_495, %view_496), kwargs = {})
%_unsafe_view_210 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%bmm_42, [2, 4096, 512]), kwargs = {})
%_param_constant584 : [#users=1] = get_attr[target=_param_constant584]
%add_293 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%_unsafe_view_210, %_param_constant584), kwargs = {})
%convert_element_type_288 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_293, torch.float32), kwargs = {})
%var_mean_93 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_288, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_186 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_93, 0), kwargs = {})
%getitem_187 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_93, 1), kwargs = {})
%add_294 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_186, 1e-05), kwargs = {})
%rsqrt_93 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_294,), kwargs = {})
%sub_93 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_293, %getitem_187), kwargs = {})
%mul_203 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_93, %rsqrt_93), kwargs = {})
%_param_constant585 : [#users=1] = get_attr[target=_param_constant585]
%mul_204 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_203, %_param_constant585), kwargs = {})
%_param_constant586 : [#users=1] = get_attr[target=_param_constant586]
%add_295 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_204, %_param_constant586), kwargs = {})
%convert_element_type_289 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_295, torch.float16), kwargs = {})
%_param_constant587 : [#users=1] = get_attr[target=_param_constant587]
%t_189 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant587,), kwargs = {})
%view_497 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_289, [8192, 512]), kwargs = {})
%mm_84 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_497, %t_189), kwargs = {})
%_unsafe_view_211 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_84, [2, 4096, 512]), kwargs = {})
%view_498 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_211, [2, 4096, 8, 64]), kwargs = {})
%permute_141 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_498, [0, 2, 1, 3]), kwargs = {})
%clone_126 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_141,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_212 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_126, [16, 4096, 64]), kwargs = {})
%_param_constant588 : [#users=1] = get_attr[target=_param_constant588]
%t_190 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant588,), kwargs = {})
%view_499 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_85 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_499, %t_190), kwargs = {})
%_unsafe_view_213 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_85, [2, 77, 512]), kwargs = {})
%_param_constant589 : [#users=1] = get_attr[target=_param_constant589]
%t_191 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant589,), kwargs = {})
%view_500 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_86 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_500, %t_191), kwargs = {})
%_unsafe_view_214 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_86, [2, 77, 512]), kwargs = {})
%view_501 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_213, [2, 77, 8, 64]), kwargs = {})
%permute_142 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_501, [0, 2, 1, 3]), kwargs = {})
%clone_127 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_142,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_215 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_127, [16, 77, 64]), kwargs = {})
%view_502 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_214, [2, 77, 8, 64]), kwargs = {})
%permute_143 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_502, [0, 2, 1, 3]), kwargs = {})
%clone_128 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_143,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_216 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_128, [16, 77, 64]), kwargs = {})
%empty_28 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 4096, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_28 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_215, -1, -2), kwargs = {})
%baddbmm_28 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_28, %_unsafe_view_212, %transpose_28), kwargs = {beta: 0, alpha: 0.125})
%_softmax_28 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_28, -1, False), kwargs = {})
%detach_130 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_28,), kwargs = {})
%bmm_43 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_28, %_unsafe_view_216), kwargs = {})
%view_503 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_43, [2, 8, 4096, 64]), kwargs = {})
%permute_144 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_503, [0, 2, 1, 3]), kwargs = {})
%clone_129 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_144,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_217 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_129, [2, 4096, 512]), kwargs = {})
%_param_constant590 : [#users=1] = get_attr[target=_param_constant590]
%t_192 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant590,), kwargs = {})
%view_504 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_217, [8192, 512]), kwargs = {})
%_param_constant591 : [#users=1] = get_attr[target=_param_constant591]
%addmm_90 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant591, %view_504, %t_192), kwargs = {})
%view_505 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_90, [2, 4096, 512]), kwargs = {})
%add_296 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_505, %add_293), kwargs = {})
%convert_element_type_290 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_296, torch.float32), kwargs = {})
%var_mean_94 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_290, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_188 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_94, 0), kwargs = {})
%getitem_189 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_94, 1), kwargs = {})
%add_297 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_188, 1e-05), kwargs = {})
%rsqrt_94 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_297,), kwargs = {})
%sub_94 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_296, %getitem_189), kwargs = {})
%mul_205 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_94, %rsqrt_94), kwargs = {})
%_param_constant592 : [#users=1] = get_attr[target=_param_constant592]
%mul_206 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_205, %_param_constant592), kwargs = {})
%_param_constant593 : [#users=1] = get_attr[target=_param_constant593]
%add_298 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_206, %_param_constant593), kwargs = {})
%convert_element_type_291 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_298, torch.float16), kwargs = {})
%_param_constant594 : [#users=1] = get_attr[target=_param_constant594]
%t_193 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant594,), kwargs = {})
%view_506 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_291, [8192, 512]), kwargs = {})
%mm_87 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_506, %t_193), kwargs = {})
%_unsafe_view_218 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_87, [2, 4096, 512]), kwargs = {})
%view_507 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_218, [2, 4096, 8, 64]), kwargs = {})
%permute_145 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_507, [0, 2, 1, 3]), kwargs = {})
%clone_130 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_145,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_219 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_130, [16, 4096, 64]), kwargs = {})
%_param_constant595 : [#users=1] = get_attr[target=_param_constant595]
%t_194 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant595,), kwargs = {})
%view_508 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_88 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_508, %t_194), kwargs = {})
%_unsafe_view_220 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_88, [2, 77, 512]), kwargs = {})
%_param_constant596 : [#users=1] = get_attr[target=_param_constant596]
%t_195 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant596,), kwargs = {})
%view_509 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_89 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_509, %t_195), kwargs = {})
%_unsafe_view_221 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_89, [2, 77, 512]), kwargs = {})
%view_510 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_220, [2, 77, 8, 64]), kwargs = {})
%permute_146 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_510, [0, 2, 1, 3]), kwargs = {})
%clone_131 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_146,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_222 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_131, [16, 77, 64]), kwargs = {})
%view_511 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_221, [2, 77, 8, 64]), kwargs = {})
%permute_147 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_511, [0, 2, 1, 3]), kwargs = {})
%clone_132 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_147,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_223 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_132, [16, 77, 64]), kwargs = {})
%empty_29 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 4096, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_29 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_222, -1, -2), kwargs = {})
%baddbmm_29 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_29, %_unsafe_view_219, %transpose_29), kwargs = {beta: 0, alpha: 0.125})
%_softmax_29 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_29, -1, False), kwargs = {})
%detach_131 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_29,), kwargs = {})
%bmm_44 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_29, %_unsafe_view_223), kwargs = {})
%view_512 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_44, [2, 8, 4096, 64]), kwargs = {})
%permute_148 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_512, [0, 2, 1, 3]), kwargs = {})
%clone_133 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_148,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_224 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_133, [2, 4096, 512]), kwargs = {})
%_param_constant597 : [#users=1] = get_attr[target=_param_constant597]
%t_196 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant597,), kwargs = {})
%view_513 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_224, [8192, 512]), kwargs = {})
%_param_constant598 : [#users=1] = get_attr[target=_param_constant598]
%addmm_91 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant598, %view_513, %t_196), kwargs = {})
%view_514 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_91, [2, 4096, 512]), kwargs = {})
%add_299 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_514, %add_296), kwargs = {})
%convert_element_type_292 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_299, torch.float32), kwargs = {})
%var_mean_95 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_292, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_190 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_95, 0), kwargs = {})
%getitem_191 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_95, 1), kwargs = {})
%add_300 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_190, 1e-05), kwargs = {})
%rsqrt_95 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_300,), kwargs = {})
%sub_95 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_299, %getitem_191), kwargs = {})
%mul_207 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_95, %rsqrt_95), kwargs = {})
%_param_constant599 : [#users=1] = get_attr[target=_param_constant599]
%mul_208 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_207, %_param_constant599), kwargs = {})
%_param_constant600 : [#users=1] = get_attr[target=_param_constant600]
%add_301 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_208, %_param_constant600), kwargs = {})
%convert_element_type_293 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_301, torch.float16), kwargs = {})
%_param_constant601 : [#users=1] = get_attr[target=_param_constant601]
%t_197 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant601,), kwargs = {})
%view_515 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_293, [8192, 512]), kwargs = {})
%_param_constant602 : [#users=1] = get_attr[target=_param_constant602]
%addmm_92 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant602, %view_515, %t_197), kwargs = {})
%view_516 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%addmm_92, [2, 4096, 4096]), kwargs = {})
%slice_71 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_516, -1, 0, 2048), kwargs = {})
%slice_72 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_516, -1, 2048, 4096), kwargs = {})
%gelu_14 : [#users=1] = call_function[target=torch.ops.aten.gelu](args = (%slice_72,), kwargs = {})
%mul_209 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%slice_71, %gelu_14), kwargs = {})
%_param_constant603 : [#users=1] = get_attr[target=_param_constant603]
%t_198 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant603,), kwargs = {})
%view_517 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_209, [8192, 2048]), kwargs = {})
%_param_constant604 : [#users=1] = get_attr[target=_param_constant604]
%addmm_93 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant604, %view_517, %t_198), kwargs = {})
%view_518 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_93, [2, 4096, 512]), kwargs = {})
%add_302 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%view_518, %add_299), kwargs = {})
%_param_constant605 : [#users=1] = get_attr[target=_param_constant605]
%t_199 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant605,), kwargs = {})
%view_519 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%add_302, [8192, 512]), kwargs = {})
%_param_constant606 : [#users=1] = get_attr[target=_param_constant606]
%addmm_94 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant606, %view_519, %t_199), kwargs = {})
%view_520 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_94, [2, 4096, 512]), kwargs = {})
%view_521 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%view_520, [2, 64, 64, 512]), kwargs = {})
%permute_149 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_521, [0, 3, 1, 2]), kwargs = {})
%clone_134 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_149,), kwargs = {memory_format: torch.contiguous_format})
%add_303 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%clone_134, %div_18), kwargs = {})
%cat_10 : [#users=2] = call_function[target=torch.ops.aten.cat](args = ([%add_303, %convolution_5], 1), kwargs = {})
%view_522 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%cat_10, [2, 32, 24, 4096]), kwargs = {})
%convert_element_type_294 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_522, torch.float32), kwargs = {})
%var_mean_96 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_294, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_192 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_96, 0), kwargs = {})
%getitem_193 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_96, 1), kwargs = {})
%add_304 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_192, 1e-05), kwargs = {})
%rsqrt_96 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_304,), kwargs = {})
%sub_96 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_522, %getitem_193), kwargs = {})
%mul_210 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_96, %rsqrt_96), kwargs = {})
%view_523 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_210, [2, 768, 64, 64]), kwargs = {})
%_param_constant607 : [#users=1] = get_attr[target=_param_constant607]
%unsqueeze_344 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant607, 0), kwargs = {})
%unsqueeze_345 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_344, 2), kwargs = {})
%unsqueeze_346 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_345, 3), kwargs = {})
%_param_constant608 : [#users=1] = get_attr[target=_param_constant608]
%unsqueeze_347 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant608, 0), kwargs = {})
%unsqueeze_348 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_347, 2), kwargs = {})
%unsqueeze_349 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_348, 3), kwargs = {})
%mul_211 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_523, %unsqueeze_349), kwargs = {})
%add_305 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_211, %unsqueeze_346), kwargs = {})
%convert_element_type_295 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_305, torch.float16), kwargs = {})
%convert_element_type_296 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_193, torch.float16), kwargs = {})
%convert_element_type_297 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_96, torch.float16), kwargs = {})
%squeeze_204 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_296, 3), kwargs = {})
%squeeze_205 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_204, 2), kwargs = {})
%squeeze_206 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_297, 3), kwargs = {})
%squeeze_207 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_206, 2), kwargs = {})
%detach_132 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_205,), kwargs = {})
%detach_133 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_207,), kwargs = {})
%silu_55 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_295,), kwargs = {})
%_param_constant609 : [#users=1] = get_attr[target=_param_constant609]
%_param_constant610 : [#users=1] = get_attr[target=_param_constant610]
%convolution_52 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_55, %_param_constant609, %_param_constant610, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_56 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant611 : [#users=1] = get_attr[target=_param_constant611]
%t_200 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant611,), kwargs = {})
%_param_constant612 : [#users=1] = get_attr[target=_param_constant612]
%addmm_95 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant612, %silu_56, %t_200), kwargs = {})
%slice_73 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_95, 0, 0, 9223372036854775807), kwargs = {})
%slice_74 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_73, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_350 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_74, 2), kwargs = {})
%unsqueeze_351 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_350, 3), kwargs = {})
%add_306 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_52, %unsqueeze_351), kwargs = {})
%view_524 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_306, [2, 32, 16, 4096]), kwargs = {})
%convert_element_type_298 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_524, torch.float32), kwargs = {})
%var_mean_97 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_298, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_194 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_97, 0), kwargs = {})
%getitem_195 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_97, 1), kwargs = {})
%add_307 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_194, 1e-05), kwargs = {})
%rsqrt_97 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_307,), kwargs = {})
%sub_97 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_524, %getitem_195), kwargs = {})
%mul_212 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_97, %rsqrt_97), kwargs = {})
%view_525 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_212, [2, 512, 64, 64]), kwargs = {})
%_param_constant613 : [#users=1] = get_attr[target=_param_constant613]
%unsqueeze_352 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant613, 0), kwargs = {})
%unsqueeze_353 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_352, 2), kwargs = {})
%unsqueeze_354 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_353, 3), kwargs = {})
%_param_constant614 : [#users=1] = get_attr[target=_param_constant614]
%unsqueeze_355 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant614, 0), kwargs = {})
%unsqueeze_356 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_355, 2), kwargs = {})
%unsqueeze_357 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_356, 3), kwargs = {})
%mul_213 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_525, %unsqueeze_357), kwargs = {})
%add_308 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_213, %unsqueeze_354), kwargs = {})
%convert_element_type_299 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_308, torch.float16), kwargs = {})
%convert_element_type_300 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_195, torch.float16), kwargs = {})
%convert_element_type_301 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_97, torch.float16), kwargs = {})
%squeeze_208 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_300, 3), kwargs = {})
%squeeze_209 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_208, 2), kwargs = {})
%squeeze_210 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_301, 3), kwargs = {})
%squeeze_211 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_210, 2), kwargs = {})
%detach_134 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_209,), kwargs = {})
%detach_135 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_211,), kwargs = {})
%silu_57 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_299,), kwargs = {})
%_param_constant615 : [#users=1] = get_attr[target=_param_constant615]
%_param_constant616 : [#users=1] = get_attr[target=_param_constant616]
%convolution_53 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_57, %_param_constant615, %_param_constant616, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%_param_constant617 : [#users=1] = get_attr[target=_param_constant617]
%_param_constant618 : [#users=1] = get_attr[target=_param_constant618]
%convolution_54 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%cat_10, %_param_constant617, %_param_constant618, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
%add_309 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_54, %convolution_53), kwargs = {})
%div_19 : [#users=2] = call_function[target=torch.ops.aten.div](args = (%add_309, 1.0), kwargs = {})
%view_526 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%div_19, [2, 32, 16, 4096]), kwargs = {})
%convert_element_type_302 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_526, torch.float32), kwargs = {})
%var_mean_98 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_302, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_196 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_98, 0), kwargs = {})
%getitem_197 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_98, 1), kwargs = {})
%add_310 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_196, 1e-06), kwargs = {})
%rsqrt_98 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_310,), kwargs = {})
%sub_98 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_526, %getitem_197), kwargs = {})
%mul_214 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_98, %rsqrt_98), kwargs = {})
%view_527 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_214, [2, 512, 64, 64]), kwargs = {})
%_param_constant619 : [#users=1] = get_attr[target=_param_constant619]
%unsqueeze_358 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant619, 0), kwargs = {})
%unsqueeze_359 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_358, 2), kwargs = {})
%unsqueeze_360 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_359, 3), kwargs = {})
%_param_constant620 : [#users=1] = get_attr[target=_param_constant620]
%unsqueeze_361 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant620, 0), kwargs = {})
%unsqueeze_362 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_361, 2), kwargs = {})
%unsqueeze_363 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_362, 3), kwargs = {})
%mul_215 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_527, %unsqueeze_363), kwargs = {})
%add_311 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_215, %unsqueeze_360), kwargs = {})
%convert_element_type_303 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_311, torch.float16), kwargs = {})
%convert_element_type_304 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_197, torch.float16), kwargs = {})
%convert_element_type_305 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_98, torch.float16), kwargs = {})
%squeeze_212 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_304, 3), kwargs = {})
%squeeze_213 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_212, 2), kwargs = {})
%squeeze_214 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_305, 3), kwargs = {})
%squeeze_215 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_214, 2), kwargs = {})
%detach_136 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_213,), kwargs = {})
%detach_137 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_215,), kwargs = {})
%permute_150 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%convert_element_type_303, [0, 2, 3, 1]), kwargs = {})
%view_528 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%permute_150, [2, 4096, 512]), kwargs = {})
%_param_constant621 : [#users=1] = get_attr[target=_param_constant621]
%t_201 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant621,), kwargs = {})
%expand_31 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%view_528, [2, 4096, 512]), kwargs = {})
%view_529 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_31, [2, 4096, 512]), kwargs = {})
%expand_32 : [#users=1] = call_function[target=torch.ops.aten.expand](args = (%t_201, [2, 512, 512]), kwargs = {})
%view_530 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%expand_32, [2, 512, 512]), kwargs = {})
%bmm_45 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%view_529, %view_530), kwargs = {})
%_unsafe_view_225 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%bmm_45, [2, 4096, 512]), kwargs = {})
%_param_constant622 : [#users=1] = get_attr[target=_param_constant622]
%add_312 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%_unsafe_view_225, %_param_constant622), kwargs = {})
%convert_element_type_306 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_312, torch.float32), kwargs = {})
%var_mean_99 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_306, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_198 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_99, 0), kwargs = {})
%getitem_199 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_99, 1), kwargs = {})
%add_313 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_198, 1e-05), kwargs = {})
%rsqrt_99 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_313,), kwargs = {})
%sub_99 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_312, %getitem_199), kwargs = {})
%mul_216 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_99, %rsqrt_99), kwargs = {})
%_param_constant623 : [#users=1] = get_attr[target=_param_constant623]
%mul_217 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_216, %_param_constant623), kwargs = {})
%_param_constant624 : [#users=1] = get_attr[target=_param_constant624]
%add_314 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_217, %_param_constant624), kwargs = {})
%convert_element_type_307 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_314, torch.float16), kwargs = {})
%_param_constant625 : [#users=1] = get_attr[target=_param_constant625]
%t_202 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant625,), kwargs = {})
%view_531 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_307, [8192, 512]), kwargs = {})
%mm_90 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_531, %t_202), kwargs = {})
%_unsafe_view_226 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_90, [2, 4096, 512]), kwargs = {})
%view_532 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_226, [2, 4096, 8, 64]), kwargs = {})
%permute_151 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_532, [0, 2, 1, 3]), kwargs = {})
%clone_135 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_151,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_227 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_135, [16, 4096, 64]), kwargs = {})
%_param_constant626 : [#users=1] = get_attr[target=_param_constant626]
%t_203 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant626,), kwargs = {})
%view_533 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_91 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_533, %t_203), kwargs = {})
%_unsafe_view_228 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_91, [2, 77, 512]), kwargs = {})
%_param_constant627 : [#users=1] = get_attr[target=_param_constant627]
%t_204 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant627,), kwargs = {})
%view_534 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_92 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_534, %t_204), kwargs = {})
%_unsafe_view_229 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_92, [2, 77, 512]), kwargs = {})
%view_535 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_228, [2, 77, 8, 64]), kwargs = {})
%permute_152 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_535, [0, 2, 1, 3]), kwargs = {})
%clone_136 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_152,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_230 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_136, [16, 77, 64]), kwargs = {})
%view_536 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_229, [2, 77, 8, 64]), kwargs = {})
%permute_153 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_536, [0, 2, 1, 3]), kwargs = {})
%clone_137 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_153,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_231 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_137, [16, 77, 64]), kwargs = {})
%empty_30 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 4096, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_30 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_230, -1, -2), kwargs = {})
%baddbmm_30 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_30, %_unsafe_view_227, %transpose_30), kwargs = {beta: 0, alpha: 0.125})
%_softmax_30 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_30, -1, False), kwargs = {})
%detach_138 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_30,), kwargs = {})
%bmm_46 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_30, %_unsafe_view_231), kwargs = {})
%view_537 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_46, [2, 8, 4096, 64]), kwargs = {})
%permute_154 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_537, [0, 2, 1, 3]), kwargs = {})
%clone_138 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_154,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_232 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_138, [2, 4096, 512]), kwargs = {})
%_param_constant628 : [#users=1] = get_attr[target=_param_constant628]
%t_205 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant628,), kwargs = {})
%view_538 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_232, [8192, 512]), kwargs = {})
%_param_constant629 : [#users=1] = get_attr[target=_param_constant629]
%addmm_96 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant629, %view_538, %t_205), kwargs = {})
%view_539 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_96, [2, 4096, 512]), kwargs = {})
%add_315 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_539, %add_312), kwargs = {})
%convert_element_type_308 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_315, torch.float32), kwargs = {})
%var_mean_100 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_308, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_200 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_100, 0), kwargs = {})
%getitem_201 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_100, 1), kwargs = {})
%add_316 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_200, 1e-05), kwargs = {})
%rsqrt_100 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_316,), kwargs = {})
%sub_100 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_315, %getitem_201), kwargs = {})
%mul_218 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_100, %rsqrt_100), kwargs = {})
%_param_constant630 : [#users=1] = get_attr[target=_param_constant630]
%mul_219 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_218, %_param_constant630), kwargs = {})
%_param_constant631 : [#users=1] = get_attr[target=_param_constant631]
%add_317 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_219, %_param_constant631), kwargs = {})
%convert_element_type_309 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_317, torch.float16), kwargs = {})
%_param_constant632 : [#users=1] = get_attr[target=_param_constant632]
%t_206 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant632,), kwargs = {})
%view_540 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_309, [8192, 512]), kwargs = {})
%mm_93 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_540, %t_206), kwargs = {})
%_unsafe_view_233 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_93, [2, 4096, 512]), kwargs = {})
%view_541 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_233, [2, 4096, 8, 64]), kwargs = {})
%permute_155 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_541, [0, 2, 1, 3]), kwargs = {})
%clone_139 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_155,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_234 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_139, [16, 4096, 64]), kwargs = {})
%_param_constant633 : [#users=1] = get_attr[target=_param_constant633]
%t_207 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant633,), kwargs = {})
%view_542 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_94 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_542, %t_207), kwargs = {})
%_unsafe_view_235 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_94, [2, 77, 512]), kwargs = {})
%_param_constant634 : [#users=1] = get_attr[target=_param_constant634]
%t_208 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant634,), kwargs = {})
%view_543 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%arg2_1, [154, 1024]), kwargs = {})
%mm_95 : [#users=1] = call_function[target=torch.ops.aten.mm](args = (%view_543, %t_208), kwargs = {})
%_unsafe_view_236 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%mm_95, [2, 77, 512]), kwargs = {})
%view_544 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_235, [2, 77, 8, 64]), kwargs = {})
%permute_156 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_544, [0, 2, 1, 3]), kwargs = {})
%clone_140 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_156,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_237 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_140, [16, 77, 64]), kwargs = {})
%view_545 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_236, [2, 77, 8, 64]), kwargs = {})
%permute_157 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_545, [0, 2, 1, 3]), kwargs = {})
%clone_141 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_157,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_238 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_141, [16, 77, 64]), kwargs = {})
%empty_31 : [#users=1] = call_function[target=torch.ops.aten.empty](args = ([16, 4096, 77],), kwargs = {dtype: torch.float16, device: cuda:0, pin_memory: False})
%transpose_31 : [#users=1] = call_function[target=torch.ops.aten.transpose](args = (%_unsafe_view_237, -1, -2), kwargs = {})
%baddbmm_31 : [#users=1] = call_function[target=torch.ops.aten.baddbmm](args = (%empty_31, %_unsafe_view_234, %transpose_31), kwargs = {beta: 0, alpha: 0.125})
%_softmax_31 : [#users=2] = call_function[target=torch.ops.aten._softmax](args = (%baddbmm_31, -1, False), kwargs = {})
%detach_139 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%_softmax_31,), kwargs = {})
%bmm_47 : [#users=1] = call_function[target=torch.ops.aten.bmm](args = (%_softmax_31, %_unsafe_view_238), kwargs = {})
%view_546 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%bmm_47, [2, 8, 4096, 64]), kwargs = {})
%permute_158 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_546, [0, 2, 1, 3]), kwargs = {})
%clone_142 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_158,), kwargs = {memory_format: torch.contiguous_format})
%_unsafe_view_239 : [#users=1] = call_function[target=torch.ops.aten._unsafe_view](args = (%clone_142, [2, 4096, 512]), kwargs = {})
%_param_constant635 : [#users=1] = get_attr[target=_param_constant635]
%t_209 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant635,), kwargs = {})
%view_547 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%_unsafe_view_239, [8192, 512]), kwargs = {})
%_param_constant636 : [#users=1] = get_attr[target=_param_constant636]
%addmm_97 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant636, %view_547, %t_209), kwargs = {})
%view_548 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_97, [2, 4096, 512]), kwargs = {})
%add_318 : [#users=3] = call_function[target=torch.ops.aten.add](args = (%view_548, %add_315), kwargs = {})
%convert_element_type_310 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_318, torch.float32), kwargs = {})
%var_mean_101 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_310, [2]), kwargs = {correction: 0, keepdim: True})
%getitem_202 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_101, 0), kwargs = {})
%getitem_203 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_101, 1), kwargs = {})
%add_319 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_202, 1e-05), kwargs = {})
%rsqrt_101 : [#users=1] = call_function[target=torch.ops.aten.rsqrt](args = (%add_319,), kwargs = {})
%sub_101 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%add_318, %getitem_203), kwargs = {})
%mul_220 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_101, %rsqrt_101), kwargs = {})
%_param_constant637 : [#users=1] = get_attr[target=_param_constant637]
%mul_221 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%mul_220, %_param_constant637), kwargs = {})
%_param_constant638 : [#users=1] = get_attr[target=_param_constant638]
%add_320 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_221, %_param_constant638), kwargs = {})
%convert_element_type_311 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_320, torch.float16), kwargs = {})
%_param_constant639 : [#users=1] = get_attr[target=_param_constant639]
%t_210 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant639,), kwargs = {})
%view_549 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%convert_element_type_311, [8192, 512]), kwargs = {})
%_param_constant640 : [#users=1] = get_attr[target=_param_constant640]
%addmm_98 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant640, %view_549, %t_210), kwargs = {})
%view_550 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%addmm_98, [2, 4096, 4096]), kwargs = {})
%slice_75 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_550, -1, 0, 2048), kwargs = {})
%slice_76 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%view_550, -1, 2048, 4096), kwargs = {})
%gelu_15 : [#users=1] = call_function[target=torch.ops.aten.gelu](args = (%slice_76,), kwargs = {})
%mul_222 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%slice_75, %gelu_15), kwargs = {})
%_param_constant641 : [#users=1] = get_attr[target=_param_constant641]
%t_211 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant641,), kwargs = {})
%view_551 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_222, [8192, 2048]), kwargs = {})
%_param_constant642 : [#users=1] = get_attr[target=_param_constant642]
%addmm_99 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant642, %view_551, %t_211), kwargs = {})
%view_552 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_99, [2, 4096, 512]), kwargs = {})
%add_321 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%view_552, %add_318), kwargs = {})
%_param_constant643 : [#users=1] = get_attr[target=_param_constant643]
%t_212 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant643,), kwargs = {})
%view_553 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%add_321, [8192, 512]), kwargs = {})
%_param_constant644 : [#users=1] = get_attr[target=_param_constant644]
%addmm_100 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant644, %view_553, %t_212), kwargs = {})
%view_554 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%addmm_100, [2, 4096, 512]), kwargs = {})
%view_555 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%view_554, [2, 64, 64, 512]), kwargs = {})
%permute_159 : [#users=1] = call_function[target=torch.ops.aten.permute](args = (%view_555, [0, 3, 1, 2]), kwargs = {})
%clone_143 : [#users=1] = call_function[target=torch.ops.aten.clone](args = (%permute_159,), kwargs = {memory_format: torch.contiguous_format})
%add_322 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%clone_143, %div_19), kwargs = {})
%upsample_nearest2d_2 : [#users=1] = call_function[target=torch.ops.aten.upsample_nearest2d](args = (%add_322, [128, 128], 2.0, 2.0), kwargs = {})
%_param_constant645 : [#users=1] = get_attr[target=_param_constant645]
%_param_constant646 : [#users=1] = get_attr[target=_param_constant646]
%convolution_55 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%upsample_nearest2d_2, %_param_constant645, %_param_constant646, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%cat_11 : [#users=2] = call_function[target=torch.ops.aten.cat](args = ([%convolution_55, %div_2], 1), kwargs = {})
%view_556 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%cat_11, [2, 32, 24, 16384]), kwargs = {})
%convert_element_type_312 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_556, torch.float32), kwargs = {})
%var_mean_102 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_312, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_204 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_102, 0), kwargs = {})
%getitem_205 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_102, 1), kwargs = {})
%add_323 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_204, 1e-05), kwargs = {})
%rsqrt_102 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_323,), kwargs = {})
%sub_102 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_556, %getitem_205), kwargs = {})
%mul_223 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_102, %rsqrt_102), kwargs = {})
%view_557 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_223, [2, 768, 128, 128]), kwargs = {})
%_param_constant647 : [#users=1] = get_attr[target=_param_constant647]
%unsqueeze_364 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant647, 0), kwargs = {})
%unsqueeze_365 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_364, 2), kwargs = {})
%unsqueeze_366 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_365, 3), kwargs = {})
%_param_constant648 : [#users=1] = get_attr[target=_param_constant648]
%unsqueeze_367 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant648, 0), kwargs = {})
%unsqueeze_368 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_367, 2), kwargs = {})
%unsqueeze_369 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_368, 3), kwargs = {})
%mul_224 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_557, %unsqueeze_369), kwargs = {})
%add_324 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_224, %unsqueeze_366), kwargs = {})
%convert_element_type_313 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_324, torch.float16), kwargs = {})
%convert_element_type_314 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_205, torch.float16), kwargs = {})
%convert_element_type_315 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_102, torch.float16), kwargs = {})
%squeeze_216 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_314, 3), kwargs = {})
%squeeze_217 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_216, 2), kwargs = {})
%squeeze_218 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_315, 3), kwargs = {})
%squeeze_219 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_218, 2), kwargs = {})
%detach_140 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_217,), kwargs = {})
%detach_141 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_219,), kwargs = {})
%silu_58 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_313,), kwargs = {})
%_param_constant649 : [#users=1] = get_attr[target=_param_constant649]
%_param_constant650 : [#users=1] = get_attr[target=_param_constant650]
%convolution_56 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_58, %_param_constant649, %_param_constant650, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_59 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant651 : [#users=1] = get_attr[target=_param_constant651]
%t_213 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant651,), kwargs = {})
%_param_constant652 : [#users=1] = get_attr[target=_param_constant652]
%addmm_101 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant652, %silu_59, %t_213), kwargs = {})
%slice_77 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_101, 0, 0, 9223372036854775807), kwargs = {})
%slice_78 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_77, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_370 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_78, 2), kwargs = {})
%unsqueeze_371 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_370, 3), kwargs = {})
%add_325 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_56, %unsqueeze_371), kwargs = {})
%view_558 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_325, [2, 32, 8, 16384]), kwargs = {})
%convert_element_type_316 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_558, torch.float32), kwargs = {})
%var_mean_103 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_316, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_206 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_103, 0), kwargs = {})
%getitem_207 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_103, 1), kwargs = {})
%add_326 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_206, 1e-05), kwargs = {})
%rsqrt_103 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_326,), kwargs = {})
%sub_103 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_558, %getitem_207), kwargs = {})
%mul_225 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_103, %rsqrt_103), kwargs = {})
%view_559 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_225, [2, 256, 128, 128]), kwargs = {})
%_param_constant653 : [#users=1] = get_attr[target=_param_constant653]
%unsqueeze_372 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant653, 0), kwargs = {})
%unsqueeze_373 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_372, 2), kwargs = {})
%unsqueeze_374 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_373, 3), kwargs = {})
%_param_constant654 : [#users=1] = get_attr[target=_param_constant654]
%unsqueeze_375 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant654, 0), kwargs = {})
%unsqueeze_376 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_375, 2), kwargs = {})
%unsqueeze_377 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_376, 3), kwargs = {})
%mul_226 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_559, %unsqueeze_377), kwargs = {})
%add_327 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_226, %unsqueeze_374), kwargs = {})
%convert_element_type_317 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_327, torch.float16), kwargs = {})
%convert_element_type_318 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_207, torch.float16), kwargs = {})
%convert_element_type_319 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_103, torch.float16), kwargs = {})
%squeeze_220 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_318, 3), kwargs = {})
%squeeze_221 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_220, 2), kwargs = {})
%squeeze_222 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_319, 3), kwargs = {})
%squeeze_223 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_222, 2), kwargs = {})
%detach_142 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_221,), kwargs = {})
%detach_143 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_223,), kwargs = {})
%silu_60 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_317,), kwargs = {})
%_param_constant655 : [#users=1] = get_attr[target=_param_constant655]
%_param_constant656 : [#users=1] = get_attr[target=_param_constant656]
%convolution_57 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_60, %_param_constant655, %_param_constant656, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%_param_constant657 : [#users=1] = get_attr[target=_param_constant657]
%_param_constant658 : [#users=1] = get_attr[target=_param_constant658]
%convolution_58 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%cat_11, %_param_constant657, %_param_constant658, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
%add_328 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_58, %convolution_57), kwargs = {})
%div_20 : [#users=1] = call_function[target=torch.ops.aten.div](args = (%add_328, 1.0), kwargs = {})
%cat_12 : [#users=2] = call_function[target=torch.ops.aten.cat](args = ([%div_20, %div_1], 1), kwargs = {})
%view_560 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%cat_12, [2, 32, 16, 16384]), kwargs = {})
%convert_element_type_320 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_560, torch.float32), kwargs = {})
%var_mean_104 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_320, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_208 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_104, 0), kwargs = {})
%getitem_209 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_104, 1), kwargs = {})
%add_329 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_208, 1e-05), kwargs = {})
%rsqrt_104 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_329,), kwargs = {})
%sub_104 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_560, %getitem_209), kwargs = {})
%mul_227 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_104, %rsqrt_104), kwargs = {})
%view_561 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_227, [2, 512, 128, 128]), kwargs = {})
%_param_constant659 : [#users=1] = get_attr[target=_param_constant659]
%unsqueeze_378 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant659, 0), kwargs = {})
%unsqueeze_379 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_378, 2), kwargs = {})
%unsqueeze_380 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_379, 3), kwargs = {})
%_param_constant660 : [#users=1] = get_attr[target=_param_constant660]
%unsqueeze_381 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant660, 0), kwargs = {})
%unsqueeze_382 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_381, 2), kwargs = {})
%unsqueeze_383 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_382, 3), kwargs = {})
%mul_228 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_561, %unsqueeze_383), kwargs = {})
%add_330 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_228, %unsqueeze_380), kwargs = {})
%convert_element_type_321 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_330, torch.float16), kwargs = {})
%convert_element_type_322 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_209, torch.float16), kwargs = {})
%convert_element_type_323 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_104, torch.float16), kwargs = {})
%squeeze_224 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_322, 3), kwargs = {})
%squeeze_225 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_224, 2), kwargs = {})
%squeeze_226 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_323, 3), kwargs = {})
%squeeze_227 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_226, 2), kwargs = {})
%detach_144 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_225,), kwargs = {})
%detach_145 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_227,), kwargs = {})
%silu_61 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_321,), kwargs = {})
%_param_constant661 : [#users=1] = get_attr[target=_param_constant661]
%_param_constant662 : [#users=1] = get_attr[target=_param_constant662]
%convolution_59 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_61, %_param_constant661, %_param_constant662, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_62 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant663 : [#users=1] = get_attr[target=_param_constant663]
%t_214 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant663,), kwargs = {})
%_param_constant664 : [#users=1] = get_attr[target=_param_constant664]
%addmm_102 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant664, %silu_62, %t_214), kwargs = {})
%slice_79 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_102, 0, 0, 9223372036854775807), kwargs = {})
%slice_80 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_79, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_384 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_80, 2), kwargs = {})
%unsqueeze_385 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_384, 3), kwargs = {})
%add_331 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_59, %unsqueeze_385), kwargs = {})
%view_562 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_331, [2, 32, 8, 16384]), kwargs = {})
%convert_element_type_324 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_562, torch.float32), kwargs = {})
%var_mean_105 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_324, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_210 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_105, 0), kwargs = {})
%getitem_211 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_105, 1), kwargs = {})
%add_332 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_210, 1e-05), kwargs = {})
%rsqrt_105 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_332,), kwargs = {})
%sub_105 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_562, %getitem_211), kwargs = {})
%mul_229 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_105, %rsqrt_105), kwargs = {})
%view_563 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_229, [2, 256, 128, 128]), kwargs = {})
%_param_constant665 : [#users=1] = get_attr[target=_param_constant665]
%unsqueeze_386 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant665, 0), kwargs = {})
%unsqueeze_387 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_386, 2), kwargs = {})
%unsqueeze_388 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_387, 3), kwargs = {})
%_param_constant666 : [#users=1] = get_attr[target=_param_constant666]
%unsqueeze_389 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant666, 0), kwargs = {})
%unsqueeze_390 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_389, 2), kwargs = {})
%unsqueeze_391 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_390, 3), kwargs = {})
%mul_230 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_563, %unsqueeze_391), kwargs = {})
%add_333 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_230, %unsqueeze_388), kwargs = {})
%convert_element_type_325 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_333, torch.float16), kwargs = {})
%convert_element_type_326 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_211, torch.float16), kwargs = {})
%convert_element_type_327 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_105, torch.float16), kwargs = {})
%squeeze_228 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_326, 3), kwargs = {})
%squeeze_229 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_228, 2), kwargs = {})
%squeeze_230 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_327, 3), kwargs = {})
%squeeze_231 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_230, 2), kwargs = {})
%detach_146 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_229,), kwargs = {})
%detach_147 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_231,), kwargs = {})
%silu_63 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_325,), kwargs = {})
%_param_constant667 : [#users=1] = get_attr[target=_param_constant667]
%_param_constant668 : [#users=1] = get_attr[target=_param_constant668]
%convolution_60 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_63, %_param_constant667, %_param_constant668, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%_param_constant669 : [#users=1] = get_attr[target=_param_constant669]
%_param_constant670 : [#users=1] = get_attr[target=_param_constant670]
%convolution_61 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%cat_12, %_param_constant669, %_param_constant670, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
%add_334 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_61, %convolution_60), kwargs = {})
%div_21 : [#users=1] = call_function[target=torch.ops.aten.div](args = (%add_334, 1.0), kwargs = {})
%cat_13 : [#users=2] = call_function[target=torch.ops.aten.cat](args = ([%div_21, %convolution], 1), kwargs = {})
%view_564 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%cat_13, [2, 32, 16, 16384]), kwargs = {})
%convert_element_type_328 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_564, torch.float32), kwargs = {})
%var_mean_106 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_328, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_212 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_106, 0), kwargs = {})
%getitem_213 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_106, 1), kwargs = {})
%add_335 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_212, 1e-05), kwargs = {})
%rsqrt_106 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_335,), kwargs = {})
%sub_106 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_564, %getitem_213), kwargs = {})
%mul_231 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_106, %rsqrt_106), kwargs = {})
%view_565 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_231, [2, 512, 128, 128]), kwargs = {})
%_param_constant671 : [#users=1] = get_attr[target=_param_constant671]
%unsqueeze_392 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant671, 0), kwargs = {})
%unsqueeze_393 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_392, 2), kwargs = {})
%unsqueeze_394 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_393, 3), kwargs = {})
%_param_constant672 : [#users=1] = get_attr[target=_param_constant672]
%unsqueeze_395 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant672, 0), kwargs = {})
%unsqueeze_396 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_395, 2), kwargs = {})
%unsqueeze_397 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_396, 3), kwargs = {})
%mul_232 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_565, %unsqueeze_397), kwargs = {})
%add_336 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_232, %unsqueeze_394), kwargs = {})
%convert_element_type_329 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_336, torch.float16), kwargs = {})
%convert_element_type_330 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_213, torch.float16), kwargs = {})
%convert_element_type_331 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_106, torch.float16), kwargs = {})
%squeeze_232 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_330, 3), kwargs = {})
%squeeze_233 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_232, 2), kwargs = {})
%squeeze_234 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_331, 3), kwargs = {})
%squeeze_235 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_234, 2), kwargs = {})
%detach_148 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_233,), kwargs = {})
%detach_149 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_235,), kwargs = {})
%silu_64 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_329,), kwargs = {})
%_param_constant673 : [#users=1] = get_attr[target=_param_constant673]
%_param_constant674 : [#users=1] = get_attr[target=_param_constant674]
%convolution_62 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_64, %_param_constant673, %_param_constant674, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%silu_65 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%add,), kwargs = {})
%_param_constant675 : [#users=1] = get_attr[target=_param_constant675]
%t_215 : [#users=1] = call_function[target=torch.ops.aten.t](args = (%_param_constant675,), kwargs = {})
%_param_constant676 : [#users=1] = get_attr[target=_param_constant676]
%addmm_103 : [#users=1] = call_function[target=torch.ops.aten.addmm](args = (%_param_constant676, %silu_65, %t_215), kwargs = {})
%slice_81 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%addmm_103, 0, 0, 9223372036854775807), kwargs = {})
%slice_82 : [#users=1] = call_function[target=torch.ops.aten.slice](args = (%slice_81, 1, 0, 9223372036854775807), kwargs = {})
%unsqueeze_398 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%slice_82, 2), kwargs = {})
%unsqueeze_399 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_398, 3), kwargs = {})
%add_337 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_62, %unsqueeze_399), kwargs = {})
%view_566 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%add_337, [2, 32, 8, 16384]), kwargs = {})
%convert_element_type_332 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_566, torch.float32), kwargs = {})
%var_mean_107 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_332, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_214 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_107, 0), kwargs = {})
%getitem_215 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_107, 1), kwargs = {})
%add_338 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_214, 1e-05), kwargs = {})
%rsqrt_107 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_338,), kwargs = {})
%sub_107 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_566, %getitem_215), kwargs = {})
%mul_233 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_107, %rsqrt_107), kwargs = {})
%view_567 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_233, [2, 256, 128, 128]), kwargs = {})
%_param_constant677 : [#users=1] = get_attr[target=_param_constant677]
%unsqueeze_400 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant677, 0), kwargs = {})
%unsqueeze_401 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_400, 2), kwargs = {})
%unsqueeze_402 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_401, 3), kwargs = {})
%_param_constant678 : [#users=1] = get_attr[target=_param_constant678]
%unsqueeze_403 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant678, 0), kwargs = {})
%unsqueeze_404 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_403, 2), kwargs = {})
%unsqueeze_405 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_404, 3), kwargs = {})
%mul_234 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_567, %unsqueeze_405), kwargs = {})
%add_339 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_234, %unsqueeze_402), kwargs = {})
%convert_element_type_333 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_339, torch.float16), kwargs = {})
%convert_element_type_334 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_215, torch.float16), kwargs = {})
%convert_element_type_335 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_107, torch.float16), kwargs = {})
%squeeze_236 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_334, 3), kwargs = {})
%squeeze_237 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_236, 2), kwargs = {})
%squeeze_238 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_335, 3), kwargs = {})
%squeeze_239 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_238, 2), kwargs = {})
%detach_150 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_237,), kwargs = {})
%detach_151 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_239,), kwargs = {})
%silu_66 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_333,), kwargs = {})
%_param_constant679 : [#users=1] = get_attr[target=_param_constant679]
%_param_constant680 : [#users=1] = get_attr[target=_param_constant680]
%convolution_63 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_66, %_param_constant679, %_param_constant680, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
%_param_constant681 : [#users=1] = get_attr[target=_param_constant681]
%_param_constant682 : [#users=1] = get_attr[target=_param_constant682]
%convolution_64 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%cat_13, %_param_constant681, %_param_constant682, [1, 1], [0, 0], [1, 1], False, [0, 0], 1), kwargs = {})
%add_340 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%convolution_64, %convolution_63), kwargs = {})
%div_22 : [#users=1] = call_function[target=torch.ops.aten.div](args = (%add_340, 1.0), kwargs = {})
%view_568 : [#users=2] = call_function[target=torch.ops.aten.view](args = (%div_22, [2, 32, 8, 16384]), kwargs = {})
%convert_element_type_336 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%view_568, torch.float32), kwargs = {})
%var_mean_108 : [#users=2] = call_function[target=torch.ops.aten.var_mean](args = (%convert_element_type_336, [2, 3]), kwargs = {correction: 0, keepdim: True})
%getitem_216 : [#users=1] = call_function[target=operator.getitem](args = (%var_mean_108, 0), kwargs = {})
%getitem_217 : [#users=2] = call_function[target=operator.getitem](args = (%var_mean_108, 1), kwargs = {})
%add_341 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%getitem_216, 1e-05), kwargs = {})
%rsqrt_108 : [#users=2] = call_function[target=torch.ops.aten.rsqrt](args = (%add_341,), kwargs = {})
%sub_108 : [#users=1] = call_function[target=torch.ops.aten.sub](args = (%view_568, %getitem_217), kwargs = {})
%mul_235 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%sub_108, %rsqrt_108), kwargs = {})
%view_569 : [#users=1] = call_function[target=torch.ops.aten.view](args = (%mul_235, [2, 256, 128, 128]), kwargs = {})
%_param_constant683 : [#users=1] = get_attr[target=_param_constant683]
%unsqueeze_406 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant683, 0), kwargs = {})
%unsqueeze_407 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_406, 2), kwargs = {})
%unsqueeze_408 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_407, 3), kwargs = {})
%_param_constant684 : [#users=1] = get_attr[target=_param_constant684]
%unsqueeze_409 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%_param_constant684, 0), kwargs = {})
%unsqueeze_410 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_409, 2), kwargs = {})
%unsqueeze_411 : [#users=1] = call_function[target=torch.ops.aten.unsqueeze](args = (%unsqueeze_410, 3), kwargs = {})
%mul_236 : [#users=1] = call_function[target=torch.ops.aten.mul](args = (%view_569, %unsqueeze_411), kwargs = {})
%add_342 : [#users=1] = call_function[target=torch.ops.aten.add](args = (%mul_236, %unsqueeze_408), kwargs = {})
%convert_element_type_337 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%add_342, torch.float16), kwargs = {})
%convert_element_type_338 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%getitem_217, torch.float16), kwargs = {})
%convert_element_type_339 : [#users=1] = call_function[target=torch.ops.prims.convert_element_type](args = (%rsqrt_108, torch.float16), kwargs = {})
%squeeze_240 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_338, 3), kwargs = {})
%squeeze_241 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_240, 2), kwargs = {})
%squeeze_242 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%convert_element_type_339, 3), kwargs = {})
%squeeze_243 : [#users=1] = call_function[target=torch.ops.aten.squeeze](args = (%squeeze_242, 2), kwargs = {})
%detach_152 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_241,), kwargs = {})
%detach_153 : [#users=0] = call_function[target=torch.ops.aten.detach](args = (%squeeze_243,), kwargs = {})
%silu_67 : [#users=1] = call_function[target=torch.ops.aten.silu](args = (%convert_element_type_337,), kwargs = {})
%_param_constant685 : [#users=1] = get_attr[target=_param_constant685]
%_param_constant686 : [#users=1] = get_attr[target=_param_constant686]
%convolution_65 : [#users=1] = call_function[target=torch.ops.aten.convolution](args = (%silu_67, %_param_constant685, %_param_constant686, [1, 1], [1, 1], [1, 1], False, [0, 0], 1), kwargs = {})
return convolution_65
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment