gpt2.mlir

module attributes {torch.debug_module_name = "GraphModule"} {
  func private @__torch__.torch.fx.graph_module.___torch_mangle_0.GraphModule.forward(%arg0: !torch.nn.Module<"__torch__.torch.fx.graph_module.___torch_mangle_0.GraphModule">, %arg1: !torch.tensor {torch.type_bound = !torch.vtensor<[50257,768],f32>}, %arg2: !torch.tensor {torch.type_bound = !torch.vtensor<[2304],f32>}, %arg3: !torch.tensor {torch.type_bound = !torch.vtensor<[768,2304],f32>}, %arg4: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg5: !torch.tensor {torch.type_bound = !torch.vtensor<[768,768],f32>}, %arg6: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg7: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg8: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg9: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg10: !torch.tensor {torch.type_bound = !torch.vtensor<[3072],f32>}, %arg11: !torch.tensor {torch.type_bound = !torch.vtensor<[768,3072],f32>}, %arg12: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg13: !torch.tensor {torch.type_bound = !torch.vtensor<[3072,768],f32>}, %arg14: !torch.tensor {torch.type_bound = !torch.vtensor<[2304],f32>}, %arg15: !torch.tensor {torch.type_bound = !torch.vtensor<[768,2304],f32>}, %arg16: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg17: !torch.tensor {torch.type_bound = !torch.vtensor<[768,768],f32>}, %arg18: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg19: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg20: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg21: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg22: !torch.tensor {torch.type_bound = !torch.vtensor<[3072],f32>}, %arg23: !torch.tensor {torch.type_bound = !torch.vtensor<[768,3072],f32>}, %arg24: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg25: !torch.tensor {torch.type_bound = !torch.vtensor<[3072,768],f32>}, %arg26: !torch.tensor {torch.type_bound = !torch.vtensor<[2304],f32>}, %arg27: !torch.tensor {torch.type_bound = !torch.vtensor<[768,2304],f32>}, %arg28: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg29: !torch.tensor {torch.type_bound = !torch.vtensor<[768,768],f32>}, %arg30: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg31: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg32: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg33: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg34: !torch.tensor {torch.type_bound = !torch.vtensor<[3072],f32>}, %arg35: !torch.tensor {torch.type_bound = !torch.vtensor<[768,3072],f32>}, %arg36: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg37: !torch.tensor {torch.type_bound = !torch.vtensor<[3072,768],f32>}, %arg38: !torch.tensor {torch.type_bound = !torch.vtensor<[2304],f32>}, %arg39: !torch.tensor {torch.type_bound = !torch.vtensor<[768,2304],f32>}, %arg40: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg41: !torch.tensor {torch.type_bound = !torch.vtensor<[768,768],f32>}, %arg42: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg43: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg44: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg45: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg46: !torch.tensor {torch.type_bound = !torch.vtensor<[3072],f32>}, %arg47: !torch.tensor {torch.type_bound = !torch.vtensor<[768,3072],f32>}, %arg48: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg49: !torch.tensor {torch.type_bound = !torch.vtensor<[3072,768],f32>}, %arg50: !torch.tensor {torch.type_bound = !torch.vtensor<[2304],f32>}, %arg51: !torch.tensor {torch.type_bound = !torch.vtensor<[768,2304],f32>}, %arg52: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg53: !torch.tensor {torch.type_bound = !torch.vtensor<[768,768],f32>}, %arg54: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg55: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg56: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg57: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg58: !torch.tensor {torch.type_bound = !torch.vtensor<[3072],f32>}, %arg59: !torch.tensor {torch.type_bound = !torch.vtensor<[768,3072],f32>}, %arg60: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg61: !torch.tensor {torch.type_bound = !torch.vtensor<[3072,768],f32>}, %arg62: !torch.tensor {torch.type_bound = !torch.vtensor<[2304],f32>}, %arg63: !torch.tensor {torch.type_bound = !torch.vtensor<[768,2304],f32>}, %arg64: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg65: !torch.tensor {torch.type_bound = !torch.vtensor<[768,768],f32>}, %arg66: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg67: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg68: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg69: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg70: !torch.tensor {torch.type_bound = !torch.vtensor<[3072],f32>}, %arg71: !torch.tensor {torch.type_bound = !torch.vtensor<[768,3072],f32>}, %arg72: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg73: !torch.tensor {torch.type_bound = !torch.vtensor<[3072,768],f32>}, %arg74: !torch.tensor {torch.type_bound = !torch.vtensor<[2304],f32>}, %arg75: !torch.tensor {torch.type_bound = !torch.vtensor<[768,2304],f32>}, %arg76: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg77: !torch.tensor {torch.type_bound = !torch.vtensor<[768,768],f32>}, %arg78: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg79: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg80: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg81: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg82: !torch.tensor {torch.type_bound = !torch.vtensor<[3072],f32>}, %arg83: !torch.tensor {torch.type_bound = !torch.vtensor<[768,3072],f32>}, %arg84: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg85: !torch.tensor {torch.type_bound = !torch.vtensor<[3072,768],f32>}, %arg86: !torch.tensor {torch.type_bound = !torch.vtensor<[2304],f32>}, %arg87: !torch.tensor {torch.type_bound = !torch.vtensor<[768,2304],f32>}, %arg88: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg89: !torch.tensor {torch.type_bound = !torch.vtensor<[768,768],f32>}, %arg90: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg91: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg92: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg93: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg94: !torch.tensor {torch.type_bound = !torch.vtensor<[3072],f32>}, %arg95: !torch.tensor {torch.type_bound = !torch.vtensor<[768,3072],f32>}, %arg96: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg97: !torch.tensor {torch.type_bound = !torch.vtensor<[3072,768],f32>}, %arg98: !torch.tensor {torch.type_bound = !torch.vtensor<[2304],f32>}, %arg99: !torch.tensor {torch.type_bound = !torch.vtensor<[768,2304],f32>}, %arg100: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg101: !torch.tensor {torch.type_bound = !torch.vtensor<[768,768],f32>}, %arg102: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg103: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg104: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg105: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg106: !torch.tensor {torch.type_bound = !torch.vtensor<[3072],f32>}, %arg107: !torch.tensor {torch.type_bound = !torch.vtensor<[768,3072],f32>}, %arg108: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg109: !torch.tensor {torch.type_bound = !torch.vtensor<[3072,768],f32>}, %arg110: !torch.tensor {torch.type_bound = !torch.vtensor<[2304],f32>}, %arg111: !torch.tensor {torch.type_bound = !torch.vtensor<[768,2304],f32>}, %arg112: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg113: !torch.tensor {torch.type_bound = !torch.vtensor<[768,768],f32>}, %arg114: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg115: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg116: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg117: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg118: !torch.tensor {torch.type_bound = !torch.vtensor<[3072],f32>}, %arg119: !torch.tensor {torch.type_bound = !torch.vtensor<[768,3072],f32>}, %arg120: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg121: !torch.tensor {torch.type_bound = !torch.vtensor<[3072,768],f32>}, %arg122: !torch.tensor {torch.type_bound = !torch.vtensor<[2304],f32>}, %arg123: !torch.tensor {torch.type_bound = !torch.vtensor<[768,2304],f32>}, %arg124: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg125: !torch.tensor {torch.type_bound = !torch.vtensor<[768,768],f32>}, %arg126: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg127: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg128: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg129: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg130: !torch.tensor {torch.type_bound = !torch.vtensor<[3072],f32>}, %arg131: !torch.tensor {torch.type_bound = !torch.vtensor<[768,3072],f32>}, %arg132: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg133: !torch.tensor {torch.type_bound = !torch.vtensor<[3072,768],f32>}, %arg134: !torch.tensor {torch.type_bound = !torch.vtensor<[2304],f32>}, %arg135: !torch.tensor {torch.type_bound = !torch.vtensor<[768,2304],f32>}, %arg136: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg137: !torch.tensor {torch.type_bound = !torch.vtensor<[768,768],f32>}, %arg138: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg139: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg140: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg141: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg142: !torch.tensor {torch.type_bound = !torch.vtensor<[3072],f32>}, %arg143: !torch.tensor {torch.type_bound = !torch.vtensor<[768,3072],f32>}, %arg144: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg145: !torch.tensor {torch.type_bound = !torch.vtensor<[3072,768],f32>}, %arg146: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg147: !torch.tensor {torch.type_bound = !torch.vtensor<[768],f32>}, %arg148: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,768],f32>}, %arg149: !torch.tensor {torch.type_bound = !torch.vtensor<[50257,768],f32>}, %arg150: !torch.tensor {torch.type_bound = !torch.vtensor<[1,1,1024,1024],ui8>}, %arg151: !torch.tensor {torch.type_bound = !torch.vtensor<[],f32>}, %arg152: !torch.tensor {torch.type_bound = !torch.vtensor<[1,1,1024,1024],ui8>}, %arg153: !torch.tensor {torch.type_bound = !torch.vtensor<[],f32>}, %arg154: !torch.tensor {torch.type_bound = !torch.vtensor<[1,1,1024,1024],ui8>}, %arg155: !torch.tensor {torch.type_bound = !torch.vtensor<[],f32>}, %arg156: !torch.tensor {torch.type_bound = !torch.vtensor<[1,1,1024,1024],ui8>}, %arg157: !torch.tensor {torch.type_bound = !torch.vtensor<[],f32>}, %arg158: !torch.tensor {torch.type_bound = !torch.vtensor<[1,1,1024,1024],ui8>}, %arg159: !torch.tensor {torch.type_bound = !torch.vtensor<[],f32>}, %arg160: !torch.tensor {torch.type_bound = !torch.vtensor<[1,1,1024,1024],ui8>}, %arg161: !torch.tensor {torch.type_bound = !torch.vtensor<[],f32>}, %arg162: !torch.tensor {torch.type_bound = !torch.vtensor<[1,1,1024,1024],ui8>}, %arg163: !torch.tensor {torch.type_bound = !torch.vtensor<[],f32>}, %arg164: !torch.tensor {torch.type_bound = !torch.vtensor<[1,1,1024,1024],ui8>}, %arg165: !torch.tensor {torch.type_bound = !torch.vtensor<[],f32>}, %arg166: !torch.tensor {torch.type_bound = !torch.vtensor<[1,1,1024,1024],ui8>}, %arg167: !torch.tensor {torch.type_bound = !torch.vtensor<[],f32>}, %arg168: !torch.tensor {torch.type_bound = !torch.vtensor<[1,1,1024,1024],ui8>}, %arg169: !torch.tensor {torch.type_bound = !torch.vtensor<[],f32>}, %arg170: !torch.tensor {torch.type_bound = !torch.vtensor<[1,1,1024,1024],ui8>}, %arg171: !torch.tensor {torch.type_bound = !torch.vtensor<[],f32>}, %arg172: !torch.tensor {torch.type_bound = !torch.vtensor<[1,1,1024,1024],ui8>}, %arg173: !torch.tensor {torch.type_bound = !torch.vtensor<[],f32>}, %arg174: !torch.tensor {torch.type_bound = !torch.vtensor<[4,512],si64>}, %arg175: !torch.tensor {torch.type_bound = !torch.vtensor<[4,512],si64>}) -> !torch.list<!torch.tensor> {
    %1 = torch.tensor.literal(dense<1.000000e+00> : tensor<f64>) : !torch.tensor<[],f64>
    %2 = torch.tensor.literal(dense<0.79788456080286541> : tensor<f64>) : !torch.tensor<[],f64>
    %3 = torch.tensor.literal(dense<4.471500e-02> : tensor<f64>) : !torch.tensor<[],f64>
    %4 = torch.tensor.literal(dense<5.000000e-01> : tensor<f64>) : !torch.tensor<[],f64>
    %none = torch.constant.none
    %5 = torch.tensor.literal(dense<8.000000e+00> : tensor<f64>) : !torch.tensor<[],f64>
    %6 = torch.tensor.literal(dense<"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tensor<1x512xsi64>) : !torch.tensor<[1,512],si64>
    %false = torch.constant.bool false
    %int-1 = torch.constant.int -1
    %int512 = torch.constant.int 512
    %int768 = torch.constant.int 768
    %float1.000000e-05 = torch.constant.float 1.000000e-05
    %int4 = torch.constant.int 4
    %int2304 = torch.constant.int 2304
    %int2 = torch.constant.int 2
    %int0 = torch.constant.int 0
    %int1 = torch.constant.int 1
    %int12 = torch.constant.int 12
    %int64 = torch.constant.int 64
    %int3 = torch.constant.int 3
    %int-2 = torch.constant.int -2
    %int48 = torch.constant.int 48
    %int9223372036854775807 = torch.constant.int 9223372036854775807
    %int11 = torch.constant.int 11
    %int3072 = torch.constant.int 3072
    %float3.000000e00 = torch.constant.float 3.000000e+00
    %int2048 = torch.constant.int 2048
    %int50257 = torch.constant.int 50257
    %int-100 = torch.constant.int -100
    %7 = torch.prim.ListConstruct %int-1, %int512 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %8 = torch.aten.view %arg174, %7 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %9 = torch.aten.embedding %arg149, %8, %int-1, %false, %false : !torch.tensor, !torch.tensor, !torch.int, !torch.bool, !torch.bool -> !torch.tensor
    %10 = torch.aten.embedding %arg148, %6, %int-1, %false, %false : !torch.tensor, !torch.tensor<[1,512],si64>, !torch.int, !torch.bool, !torch.bool -> !torch.tensor
    %11 = torch.aten.add.Tensor %9, %10, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %12 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0, %result1, %result2 = torch.aten.native_layer_norm %11, %12, %arg7, %arg6, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %13 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %14 = torch.aten.view %result0, %13 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %15 = torch.aten.addmm %arg2, %14, %arg3, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %16 = torch.prim.ListConstruct %int4, %int512, %int2304 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %17 = torch.aten.view %15, %16 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %18 = torch.operator "aten.split.Tensor"(%17, %int768, %int2) : (!torch.tensor, !torch.int, !torch.int) -> !torch.list<!torch.tensor>
    %19 = torch.aten.__getitem__.t %18, %int0 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %20 = torch.aten.__getitem__.t %18, %int1 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %21 = torch.aten.__getitem__.t %18, %int2 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %22 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %23 = torch.aten.view %19, %22 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %24 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %25 = torch.aten.permute %23, %24 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %26 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %27 = torch.aten.view %20, %26 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %28 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %29 = torch.aten.permute %27, %28 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %30 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %31 = torch.aten.view %21, %30 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %32 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %33 = torch.aten.permute %31, %32 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %34 = torch.aten.transpose.int %29, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %35 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %36 = torch.aten.expand %25, %35, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %37 = torch.aten.clone %36, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %38 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %39 = torch.aten._unsafe_view %37, %38 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %40 = torch.prim.ListConstruct %int4, %int12, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %41 = torch.aten.expand %34, %40, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %42 = torch.aten.clone %41, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %43 = torch.prim.ListConstruct %int48, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %44 = torch.aten._unsafe_view %42, %43 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %45 = torch.aten.bmm %39, %44 : !torch.tensor, !torch.tensor -> !torch.tensor
    %46 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %47 = torch.aten._unsafe_view %45, %46 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %48 = torch.aten.div.Tensor %47, %5 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %49 = torch.aten.slice.Tensor %arg150, %int0, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %50 = torch.aten.slice.Tensor %49, %int1, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %51 = torch.aten.slice.Tensor %50, %int2, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %52 = torch.aten.slice.Tensor %51, %int3, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %53 = torch.operator "aten._to_copy"(%52, %int11, %none, %none, %none, %false, %none) : (!torch.tensor, !torch.int, !torch.none, !torch.none, !torch.none, !torch.bool, !torch.none) -> !torch.tensor
    %54 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %55 = torch.aten.expand %arg151, %54, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %56 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %57 = torch.aten.expand %48, %56, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %58 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %59 = torch.aten.expand %53, %58, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %60 = torch.operator "aten._s_where"(%59, %57, %55) : (!torch.tensor, !torch.tensor, !torch.tensor) -> !torch.tensor
    %61 = torch.aten._softmax %60, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %62 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %63 = torch.aten.expand %61, %62, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %64 = torch.prim.ListConstruct %int48, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %65 = torch.aten.view %63, %64 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %66 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %67 = torch.aten.expand %33, %66, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %68 = torch.aten.clone %67, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %69 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %70 = torch.aten._unsafe_view %68, %69 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %71 = torch.aten.bmm %65, %70 : !torch.tensor, !torch.tensor -> !torch.tensor
    %72 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %73 = torch.aten._unsafe_view %71, %72 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %74 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %75 = torch.aten.permute %73, %74 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %76 = torch.aten.clone %75, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %77 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %78 = torch.aten.view %76, %77 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %79 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %80 = torch.aten.view %78, %79 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %81 = torch.aten.addmm %arg4, %80, %arg5, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %82 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %83 = torch.aten.view %81, %82 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %84 = torch.aten.add.Tensor %83, %11, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %85 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_0, %result1_1, %result2_2 = torch.aten.native_layer_norm %84, %85, %arg9, %arg8, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %86 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %87 = torch.aten.view %result0_0, %86 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %88 = torch.aten.addmm %arg10, %87, %arg11, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %89 = torch.prim.ListConstruct %int4, %int512, %int3072 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %90 = torch.aten.view %88, %89 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %91 = torch.aten.mul.Tensor %90, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %92 = torch.aten.pow.Tensor_Scalar %90, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %93 = torch.aten.mul.Tensor %92, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %94 = torch.aten.add.Tensor %90, %93, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %95 = torch.aten.mul.Tensor %94, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %96 = torch.aten.tanh %95 : !torch.tensor -> !torch.tensor
    %97 = torch.aten.add.Tensor %96, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %98 = torch.aten.mul.Tensor %91, %97 : !torch.tensor, !torch.tensor -> !torch.tensor
    %99 = torch.prim.ListConstruct %int-1, %int3072 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %100 = torch.aten.view %98, %99 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %101 = torch.aten.addmm %arg12, %100, %arg13, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %102 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %103 = torch.aten.view %101, %102 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %104 = torch.aten.add.Tensor %84, %103, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %105 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_3, %result1_4, %result2_5 = torch.aten.native_layer_norm %104, %105, %arg19, %arg18, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %106 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %107 = torch.aten.view %result0_3, %106 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %108 = torch.aten.addmm %arg14, %107, %arg15, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %109 = torch.prim.ListConstruct %int4, %int512, %int2304 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %110 = torch.aten.view %108, %109 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %111 = torch.operator "aten.split.Tensor"(%110, %int768, %int2) : (!torch.tensor, !torch.int, !torch.int) -> !torch.list<!torch.tensor>
    %112 = torch.aten.__getitem__.t %111, %int0 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %113 = torch.aten.__getitem__.t %111, %int1 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %114 = torch.aten.__getitem__.t %111, %int2 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %115 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %116 = torch.aten.view %112, %115 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %117 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %118 = torch.aten.permute %116, %117 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %119 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %120 = torch.aten.view %113, %119 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %121 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %122 = torch.aten.permute %120, %121 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %123 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %124 = torch.aten.view %114, %123 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %125 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %126 = torch.aten.permute %124, %125 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %127 = torch.aten.transpose.int %122, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %128 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %129 = torch.aten.expand %118, %128, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %130 = torch.aten.clone %129, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %131 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %132 = torch.aten._unsafe_view %130, %131 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %133 = torch.prim.ListConstruct %int4, %int12, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %134 = torch.aten.expand %127, %133, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %135 = torch.aten.clone %134, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %136 = torch.prim.ListConstruct %int48, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %137 = torch.aten._unsafe_view %135, %136 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %138 = torch.aten.bmm %132, %137 : !torch.tensor, !torch.tensor -> !torch.tensor
    %139 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %140 = torch.aten._unsafe_view %138, %139 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %141 = torch.aten.div.Tensor %140, %5 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %142 = torch.aten.slice.Tensor %arg152, %int0, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %143 = torch.aten.slice.Tensor %142, %int1, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %144 = torch.aten.slice.Tensor %143, %int2, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %145 = torch.aten.slice.Tensor %144, %int3, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %146 = torch.operator "aten._to_copy"(%145, %int11, %none, %none, %none, %false, %none) : (!torch.tensor, !torch.int, !torch.none, !torch.none, !torch.none, !torch.bool, !torch.none) -> !torch.tensor
    %147 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %148 = torch.aten.expand %arg153, %147, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %149 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %150 = torch.aten.expand %141, %149, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %151 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %152 = torch.aten.expand %146, %151, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %153 = torch.operator "aten._s_where"(%152, %150, %148) : (!torch.tensor, !torch.tensor, !torch.tensor) -> !torch.tensor
    %154 = torch.aten._softmax %153, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %155 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %156 = torch.aten.expand %154, %155, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %157 = torch.prim.ListConstruct %int48, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %158 = torch.aten.view %156, %157 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %159 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %160 = torch.aten.expand %126, %159, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %161 = torch.aten.clone %160, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %162 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %163 = torch.aten._unsafe_view %161, %162 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %164 = torch.aten.bmm %158, %163 : !torch.tensor, !torch.tensor -> !torch.tensor
    %165 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %166 = torch.aten._unsafe_view %164, %165 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %167 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %168 = torch.aten.permute %166, %167 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %169 = torch.aten.clone %168, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %170 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %171 = torch.aten.view %169, %170 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %172 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %173 = torch.aten.view %171, %172 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %174 = torch.aten.addmm %arg16, %173, %arg17, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %175 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %176 = torch.aten.view %174, %175 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %177 = torch.aten.add.Tensor %176, %104, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %178 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_6, %result1_7, %result2_8 = torch.aten.native_layer_norm %177, %178, %arg21, %arg20, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %179 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %180 = torch.aten.view %result0_6, %179 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %181 = torch.aten.addmm %arg22, %180, %arg23, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %182 = torch.prim.ListConstruct %int4, %int512, %int3072 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %183 = torch.aten.view %181, %182 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %184 = torch.aten.mul.Tensor %183, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %185 = torch.aten.pow.Tensor_Scalar %183, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %186 = torch.aten.mul.Tensor %185, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %187 = torch.aten.add.Tensor %183, %186, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %188 = torch.aten.mul.Tensor %187, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %189 = torch.aten.tanh %188 : !torch.tensor -> !torch.tensor
    %190 = torch.aten.add.Tensor %189, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %191 = torch.aten.mul.Tensor %184, %190 : !torch.tensor, !torch.tensor -> !torch.tensor
    %192 = torch.prim.ListConstruct %int-1, %int3072 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %193 = torch.aten.view %191, %192 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %194 = torch.aten.addmm %arg24, %193, %arg25, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %195 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %196 = torch.aten.view %194, %195 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %197 = torch.aten.add.Tensor %177, %196, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %198 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_9, %result1_10, %result2_11 = torch.aten.native_layer_norm %197, %198, %arg55, %arg54, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %199 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %200 = torch.aten.view %result0_9, %199 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %201 = torch.aten.addmm %arg50, %200, %arg51, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %202 = torch.prim.ListConstruct %int4, %int512, %int2304 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %203 = torch.aten.view %201, %202 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %204 = torch.operator "aten.split.Tensor"(%203, %int768, %int2) : (!torch.tensor, !torch.int, !torch.int) -> !torch.list<!torch.tensor>
    %205 = torch.aten.__getitem__.t %204, %int0 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %206 = torch.aten.__getitem__.t %204, %int1 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %207 = torch.aten.__getitem__.t %204, %int2 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %208 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %209 = torch.aten.view %205, %208 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %210 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %211 = torch.aten.permute %209, %210 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %212 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %213 = torch.aten.view %206, %212 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %214 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %215 = torch.aten.permute %213, %214 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %216 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %217 = torch.aten.view %207, %216 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %218 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %219 = torch.aten.permute %217, %218 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %220 = torch.aten.transpose.int %215, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %221 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %222 = torch.aten.expand %211, %221, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %223 = torch.aten.clone %222, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %224 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %225 = torch.aten._unsafe_view %223, %224 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %226 = torch.prim.ListConstruct %int4, %int12, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %227 = torch.aten.expand %220, %226, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %228 = torch.aten.clone %227, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %229 = torch.prim.ListConstruct %int48, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %230 = torch.aten._unsafe_view %228, %229 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %231 = torch.aten.bmm %225, %230 : !torch.tensor, !torch.tensor -> !torch.tensor
    %232 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %233 = torch.aten._unsafe_view %231, %232 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %234 = torch.aten.div.Tensor %233, %5 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %235 = torch.aten.slice.Tensor %arg158, %int0, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %236 = torch.aten.slice.Tensor %235, %int1, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %237 = torch.aten.slice.Tensor %236, %int2, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %238 = torch.aten.slice.Tensor %237, %int3, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %239 = torch.operator "aten._to_copy"(%238, %int11, %none, %none, %none, %false, %none) : (!torch.tensor, !torch.int, !torch.none, !torch.none, !torch.none, !torch.bool, !torch.none) -> !torch.tensor
    %240 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %241 = torch.aten.expand %arg159, %240, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %242 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %243 = torch.aten.expand %234, %242, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %244 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %245 = torch.aten.expand %239, %244, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %246 = torch.operator "aten._s_where"(%245, %243, %241) : (!torch.tensor, !torch.tensor, !torch.tensor) -> !torch.tensor
    %247 = torch.aten._softmax %246, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %248 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %249 = torch.aten.expand %247, %248, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %250 = torch.prim.ListConstruct %int48, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %251 = torch.aten.view %249, %250 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %252 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %253 = torch.aten.expand %219, %252, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %254 = torch.aten.clone %253, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %255 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %256 = torch.aten._unsafe_view %254, %255 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %257 = torch.aten.bmm %251, %256 : !torch.tensor, !torch.tensor -> !torch.tensor
    %258 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %259 = torch.aten._unsafe_view %257, %258 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %260 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %261 = torch.aten.permute %259, %260 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %262 = torch.aten.clone %261, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %263 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %264 = torch.aten.view %262, %263 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %265 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %266 = torch.aten.view %264, %265 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %267 = torch.aten.addmm %arg52, %266, %arg53, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %268 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %269 = torch.aten.view %267, %268 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %270 = torch.aten.add.Tensor %269, %197, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %271 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_12, %result1_13, %result2_14 = torch.aten.native_layer_norm %270, %271, %arg57, %arg56, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %272 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %273 = torch.aten.view %result0_12, %272 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %274 = torch.aten.addmm %arg58, %273, %arg59, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %275 = torch.prim.ListConstruct %int4, %int512, %int3072 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %276 = torch.aten.view %274, %275 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %277 = torch.aten.mul.Tensor %276, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %278 = torch.aten.pow.Tensor_Scalar %276, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %279 = torch.aten.mul.Tensor %278, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %280 = torch.aten.add.Tensor %276, %279, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %281 = torch.aten.mul.Tensor %280, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %282 = torch.aten.tanh %281 : !torch.tensor -> !torch.tensor
    %283 = torch.aten.add.Tensor %282, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %284 = torch.aten.mul.Tensor %277, %283 : !torch.tensor, !torch.tensor -> !torch.tensor
    %285 = torch.prim.ListConstruct %int-1, %int3072 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %286 = torch.aten.view %284, %285 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %287 = torch.aten.addmm %arg60, %286, %arg61, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %288 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %289 = torch.aten.view %287, %288 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %290 = torch.aten.add.Tensor %270, %289, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %291 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_15, %result1_16, %result2_17 = torch.aten.native_layer_norm %290, %291, %arg67, %arg66, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %292 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %293 = torch.aten.view %result0_15, %292 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %294 = torch.aten.addmm %arg62, %293, %arg63, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %295 = torch.prim.ListConstruct %int4, %int512, %int2304 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %296 = torch.aten.view %294, %295 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %297 = torch.operator "aten.split.Tensor"(%296, %int768, %int2) : (!torch.tensor, !torch.int, !torch.int) -> !torch.list<!torch.tensor>
    %298 = torch.aten.__getitem__.t %297, %int0 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %299 = torch.aten.__getitem__.t %297, %int1 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %300 = torch.aten.__getitem__.t %297, %int2 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %301 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %302 = torch.aten.view %298, %301 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %303 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %304 = torch.aten.permute %302, %303 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %305 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %306 = torch.aten.view %299, %305 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %307 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %308 = torch.aten.permute %306, %307 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %309 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %310 = torch.aten.view %300, %309 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %311 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %312 = torch.aten.permute %310, %311 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %313 = torch.aten.transpose.int %308, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %314 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %315 = torch.aten.expand %304, %314, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %316 = torch.aten.clone %315, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %317 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %318 = torch.aten._unsafe_view %316, %317 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %319 = torch.prim.ListConstruct %int4, %int12, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %320 = torch.aten.expand %313, %319, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %321 = torch.aten.clone %320, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %322 = torch.prim.ListConstruct %int48, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %323 = torch.aten._unsafe_view %321, %322 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %324 = torch.aten.bmm %318, %323 : !torch.tensor, !torch.tensor -> !torch.tensor
    %325 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %326 = torch.aten._unsafe_view %324, %325 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %327 = torch.aten.div.Tensor %326, %5 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %328 = torch.aten.slice.Tensor %arg160, %int0, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %329 = torch.aten.slice.Tensor %328, %int1, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %330 = torch.aten.slice.Tensor %329, %int2, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %331 = torch.aten.slice.Tensor %330, %int3, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %332 = torch.operator "aten._to_copy"(%331, %int11, %none, %none, %none, %false, %none) : (!torch.tensor, !torch.int, !torch.none, !torch.none, !torch.none, !torch.bool, !torch.none) -> !torch.tensor
    %333 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %334 = torch.aten.expand %arg161, %333, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %335 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %336 = torch.aten.expand %327, %335, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %337 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %338 = torch.aten.expand %332, %337, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %339 = torch.operator "aten._s_where"(%338, %336, %334) : (!torch.tensor, !torch.tensor, !torch.tensor) -> !torch.tensor
    %340 = torch.aten._softmax %339, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %341 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %342 = torch.aten.expand %340, %341, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %343 = torch.prim.ListConstruct %int48, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %344 = torch.aten.view %342, %343 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %345 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %346 = torch.aten.expand %312, %345, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %347 = torch.aten.clone %346, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %348 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %349 = torch.aten._unsafe_view %347, %348 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %350 = torch.aten.bmm %344, %349 : !torch.tensor, !torch.tensor -> !torch.tensor
    %351 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %352 = torch.aten._unsafe_view %350, %351 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %353 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %354 = torch.aten.permute %352, %353 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %355 = torch.aten.clone %354, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %356 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %357 = torch.aten.view %355, %356 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %358 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %359 = torch.aten.view %357, %358 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %360 = torch.aten.addmm %arg64, %359, %arg65, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %361 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %362 = torch.aten.view %360, %361 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %363 = torch.aten.add.Tensor %362, %290, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %364 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_18, %result1_19, %result2_20 = torch.aten.native_layer_norm %363, %364, %arg69, %arg68, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %365 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %366 = torch.aten.view %result0_18, %365 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %367 = torch.aten.addmm %arg70, %366, %arg71, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %368 = torch.prim.ListConstruct %int4, %int512, %int3072 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %369 = torch.aten.view %367, %368 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %370 = torch.aten.mul.Tensor %369, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %371 = torch.aten.pow.Tensor_Scalar %369, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %372 = torch.aten.mul.Tensor %371, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %373 = torch.aten.add.Tensor %369, %372, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %374 = torch.aten.mul.Tensor %373, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %375 = torch.aten.tanh %374 : !torch.tensor -> !torch.tensor
    %376 = torch.aten.add.Tensor %375, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %377 = torch.aten.mul.Tensor %370, %376 : !torch.tensor, !torch.tensor -> !torch.tensor
    %378 = torch.prim.ListConstruct %int-1, %int3072 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %379 = torch.aten.view %377, %378 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %380 = torch.aten.addmm %arg72, %379, %arg73, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %381 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %382 = torch.aten.view %380, %381 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %383 = torch.aten.add.Tensor %363, %382, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %384 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_21, %result1_22, %result2_23 = torch.aten.native_layer_norm %383, %384, %arg79, %arg78, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %385 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %386 = torch.aten.view %result0_21, %385 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %387 = torch.aten.addmm %arg74, %386, %arg75, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %388 = torch.prim.ListConstruct %int4, %int512, %int2304 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %389 = torch.aten.view %387, %388 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %390 = torch.operator "aten.split.Tensor"(%389, %int768, %int2) : (!torch.tensor, !torch.int, !torch.int) -> !torch.list<!torch.tensor>
    %391 = torch.aten.__getitem__.t %390, %int0 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %392 = torch.aten.__getitem__.t %390, %int1 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %393 = torch.aten.__getitem__.t %390, %int2 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %394 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %395 = torch.aten.view %391, %394 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %396 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %397 = torch.aten.permute %395, %396 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %398 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %399 = torch.aten.view %392, %398 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %400 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %401 = torch.aten.permute %399, %400 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %402 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %403 = torch.aten.view %393, %402 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %404 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %405 = torch.aten.permute %403, %404 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %406 = torch.aten.transpose.int %401, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %407 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %408 = torch.aten.expand %397, %407, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %409 = torch.aten.clone %408, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %410 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %411 = torch.aten._unsafe_view %409, %410 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %412 = torch.prim.ListConstruct %int4, %int12, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %413 = torch.aten.expand %406, %412, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %414 = torch.aten.clone %413, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %415 = torch.prim.ListConstruct %int48, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %416 = torch.aten._unsafe_view %414, %415 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %417 = torch.aten.bmm %411, %416 : !torch.tensor, !torch.tensor -> !torch.tensor
    %418 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %419 = torch.aten._unsafe_view %417, %418 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %420 = torch.aten.div.Tensor %419, %5 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %421 = torch.aten.slice.Tensor %arg162, %int0, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %422 = torch.aten.slice.Tensor %421, %int1, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %423 = torch.aten.slice.Tensor %422, %int2, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %424 = torch.aten.slice.Tensor %423, %int3, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %425 = torch.operator "aten._to_copy"(%424, %int11, %none, %none, %none, %false, %none) : (!torch.tensor, !torch.int, !torch.none, !torch.none, !torch.none, !torch.bool, !torch.none) -> !torch.tensor
    %426 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %427 = torch.aten.expand %arg163, %426, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %428 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %429 = torch.aten.expand %420, %428, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %430 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %431 = torch.aten.expand %425, %430, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %432 = torch.operator "aten._s_where"(%431, %429, %427) : (!torch.tensor, !torch.tensor, !torch.tensor) -> !torch.tensor
    %433 = torch.aten._softmax %432, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %434 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %435 = torch.aten.expand %433, %434, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %436 = torch.prim.ListConstruct %int48, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %437 = torch.aten.view %435, %436 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %438 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %439 = torch.aten.expand %405, %438, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %440 = torch.aten.clone %439, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %441 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %442 = torch.aten._unsafe_view %440, %441 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %443 = torch.aten.bmm %437, %442 : !torch.tensor, !torch.tensor -> !torch.tensor
    %444 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %445 = torch.aten._unsafe_view %443, %444 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %446 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %447 = torch.aten.permute %445, %446 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %448 = torch.aten.clone %447, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %449 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %450 = torch.aten.view %448, %449 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %451 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %452 = torch.aten.view %450, %451 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %453 = torch.aten.addmm %arg76, %452, %arg77, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %454 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %455 = torch.aten.view %453, %454 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %456 = torch.aten.add.Tensor %455, %383, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %457 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_24, %result1_25, %result2_26 = torch.aten.native_layer_norm %456, %457, %arg81, %arg80, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %458 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %459 = torch.aten.view %result0_24, %458 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %460 = torch.aten.addmm %arg82, %459, %arg83, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %461 = torch.prim.ListConstruct %int4, %int512, %int3072 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %462 = torch.aten.view %460, %461 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %463 = torch.aten.mul.Tensor %462, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %464 = torch.aten.pow.Tensor_Scalar %462, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %465 = torch.aten.mul.Tensor %464, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %466 = torch.aten.add.Tensor %462, %465, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %467 = torch.aten.mul.Tensor %466, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %468 = torch.aten.tanh %467 : !torch.tensor -> !torch.tensor
    %469 = torch.aten.add.Tensor %468, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %470 = torch.aten.mul.Tensor %463, %469 : !torch.tensor, !torch.tensor -> !torch.tensor
    %471 = torch.prim.ListConstruct %int-1, %int3072 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %472 = torch.aten.view %470, %471 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %473 = torch.aten.addmm %arg84, %472, %arg85, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %474 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %475 = torch.aten.view %473, %474 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %476 = torch.aten.add.Tensor %456, %475, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %477 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_27, %result1_28, %result2_29 = torch.aten.native_layer_norm %476, %477, %arg91, %arg90, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %478 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %479 = torch.aten.view %result0_27, %478 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %480 = torch.aten.addmm %arg86, %479, %arg87, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %481 = torch.prim.ListConstruct %int4, %int512, %int2304 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %482 = torch.aten.view %480, %481 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %483 = torch.operator "aten.split.Tensor"(%482, %int768, %int2) : (!torch.tensor, !torch.int, !torch.int) -> !torch.list<!torch.tensor>
    %484 = torch.aten.__getitem__.t %483, %int0 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %485 = torch.aten.__getitem__.t %483, %int1 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %486 = torch.aten.__getitem__.t %483, %int2 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %487 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %488 = torch.aten.view %484, %487 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %489 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %490 = torch.aten.permute %488, %489 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %491 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %492 = torch.aten.view %485, %491 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %493 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %494 = torch.aten.permute %492, %493 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %495 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %496 = torch.aten.view %486, %495 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %497 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %498 = torch.aten.permute %496, %497 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %499 = torch.aten.transpose.int %494, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %500 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %501 = torch.aten.expand %490, %500, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %502 = torch.aten.clone %501, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %503 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %504 = torch.aten._unsafe_view %502, %503 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %505 = torch.prim.ListConstruct %int4, %int12, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %506 = torch.aten.expand %499, %505, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %507 = torch.aten.clone %506, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %508 = torch.prim.ListConstruct %int48, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %509 = torch.aten._unsafe_view %507, %508 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %510 = torch.aten.bmm %504, %509 : !torch.tensor, !torch.tensor -> !torch.tensor
    %511 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %512 = torch.aten._unsafe_view %510, %511 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %513 = torch.aten.div.Tensor %512, %5 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %514 = torch.aten.slice.Tensor %arg164, %int0, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %515 = torch.aten.slice.Tensor %514, %int1, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %516 = torch.aten.slice.Tensor %515, %int2, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %517 = torch.aten.slice.Tensor %516, %int3, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %518 = torch.operator "aten._to_copy"(%517, %int11, %none, %none, %none, %false, %none) : (!torch.tensor, !torch.int, !torch.none, !torch.none, !torch.none, !torch.bool, !torch.none) -> !torch.tensor
    %519 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %520 = torch.aten.expand %arg165, %519, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %521 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %522 = torch.aten.expand %513, %521, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %523 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %524 = torch.aten.expand %518, %523, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %525 = torch.operator "aten._s_where"(%524, %522, %520) : (!torch.tensor, !torch.tensor, !torch.tensor) -> !torch.tensor
    %526 = torch.aten._softmax %525, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %527 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %528 = torch.aten.expand %526, %527, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %529 = torch.prim.ListConstruct %int48, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %530 = torch.aten.view %528, %529 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %531 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %532 = torch.aten.expand %498, %531, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %533 = torch.aten.clone %532, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %534 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %535 = torch.aten._unsafe_view %533, %534 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %536 = torch.aten.bmm %530, %535 : !torch.tensor, !torch.tensor -> !torch.tensor
    %537 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %538 = torch.aten._unsafe_view %536, %537 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %539 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %540 = torch.aten.permute %538, %539 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %541 = torch.aten.clone %540, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %542 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %543 = torch.aten.view %541, %542 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %544 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %545 = torch.aten.view %543, %544 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %546 = torch.aten.addmm %arg88, %545, %arg89, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %547 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %548 = torch.aten.view %546, %547 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %549 = torch.aten.add.Tensor %548, %476, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %550 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_30, %result1_31, %result2_32 = torch.aten.native_layer_norm %549, %550, %arg93, %arg92, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %551 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %552 = torch.aten.view %result0_30, %551 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %553 = torch.aten.addmm %arg94, %552, %arg95, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %554 = torch.prim.ListConstruct %int4, %int512, %int3072 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %555 = torch.aten.view %553, %554 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %556 = torch.aten.mul.Tensor %555, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %557 = torch.aten.pow.Tensor_Scalar %555, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %558 = torch.aten.mul.Tensor %557, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %559 = torch.aten.add.Tensor %555, %558, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %560 = torch.aten.mul.Tensor %559, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %561 = torch.aten.tanh %560 : !torch.tensor -> !torch.tensor
    %562 = torch.aten.add.Tensor %561, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %563 = torch.aten.mul.Tensor %556, %562 : !torch.tensor, !torch.tensor -> !torch.tensor
    %564 = torch.prim.ListConstruct %int-1, %int3072 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %565 = torch.aten.view %563, %564 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %566 = torch.aten.addmm %arg96, %565, %arg97, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %567 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %568 = torch.aten.view %566, %567 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %569 = torch.aten.add.Tensor %549, %568, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %570 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_33, %result1_34, %result2_35 = torch.aten.native_layer_norm %569, %570, %arg103, %arg102, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %571 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %572 = torch.aten.view %result0_33, %571 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %573 = torch.aten.addmm %arg98, %572, %arg99, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %574 = torch.prim.ListConstruct %int4, %int512, %int2304 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %575 = torch.aten.view %573, %574 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %576 = torch.operator "aten.split.Tensor"(%575, %int768, %int2) : (!torch.tensor, !torch.int, !torch.int) -> !torch.list<!torch.tensor>
    %577 = torch.aten.__getitem__.t %576, %int0 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %578 = torch.aten.__getitem__.t %576, %int1 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %579 = torch.aten.__getitem__.t %576, %int2 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %580 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %581 = torch.aten.view %577, %580 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %582 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %583 = torch.aten.permute %581, %582 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %584 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %585 = torch.aten.view %578, %584 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %586 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %587 = torch.aten.permute %585, %586 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %588 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %589 = torch.aten.view %579, %588 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %590 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %591 = torch.aten.permute %589, %590 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %592 = torch.aten.transpose.int %587, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %593 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %594 = torch.aten.expand %583, %593, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %595 = torch.aten.clone %594, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %596 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %597 = torch.aten._unsafe_view %595, %596 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %598 = torch.prim.ListConstruct %int4, %int12, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %599 = torch.aten.expand %592, %598, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %600 = torch.aten.clone %599, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %601 = torch.prim.ListConstruct %int48, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %602 = torch.aten._unsafe_view %600, %601 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %603 = torch.aten.bmm %597, %602 : !torch.tensor, !torch.tensor -> !torch.tensor
    %604 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %605 = torch.aten._unsafe_view %603, %604 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %606 = torch.aten.div.Tensor %605, %5 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %607 = torch.aten.slice.Tensor %arg166, %int0, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %608 = torch.aten.slice.Tensor %607, %int1, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %609 = torch.aten.slice.Tensor %608, %int2, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %610 = torch.aten.slice.Tensor %609, %int3, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %611 = torch.operator "aten._to_copy"(%610, %int11, %none, %none, %none, %false, %none) : (!torch.tensor, !torch.int, !torch.none, !torch.none, !torch.none, !torch.bool, !torch.none) -> !torch.tensor
    %612 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %613 = torch.aten.expand %arg167, %612, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %614 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %615 = torch.aten.expand %606, %614, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %616 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %617 = torch.aten.expand %611, %616, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %618 = torch.operator "aten._s_where"(%617, %615, %613) : (!torch.tensor, !torch.tensor, !torch.tensor) -> !torch.tensor
    %619 = torch.aten._softmax %618, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %620 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %621 = torch.aten.expand %619, %620, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %622 = torch.prim.ListConstruct %int48, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %623 = torch.aten.view %621, %622 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %624 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %625 = torch.aten.expand %591, %624, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %626 = torch.aten.clone %625, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %627 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %628 = torch.aten._unsafe_view %626, %627 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %629 = torch.aten.bmm %623, %628 : !torch.tensor, !torch.tensor -> !torch.tensor
    %630 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %631 = torch.aten._unsafe_view %629, %630 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %632 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %633 = torch.aten.permute %631, %632 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %634 = torch.aten.clone %633, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %635 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %636 = torch.aten.view %634, %635 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %637 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %638 = torch.aten.view %636, %637 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %639 = torch.aten.addmm %arg100, %638, %arg101, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %640 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %641 = torch.aten.view %639, %640 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %642 = torch.aten.add.Tensor %641, %569, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %643 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_36, %result1_37, %result2_38 = torch.aten.native_layer_norm %642, %643, %arg105, %arg104, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %644 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %645 = torch.aten.view %result0_36, %644 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %646 = torch.aten.addmm %arg106, %645, %arg107, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %647 = torch.prim.ListConstruct %int4, %int512, %int3072 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %648 = torch.aten.view %646, %647 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %649 = torch.aten.mul.Tensor %648, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %650 = torch.aten.pow.Tensor_Scalar %648, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %651 = torch.aten.mul.Tensor %650, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %652 = torch.aten.add.Tensor %648, %651, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %653 = torch.aten.mul.Tensor %652, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %654 = torch.aten.tanh %653 : !torch.tensor -> !torch.tensor
    %655 = torch.aten.add.Tensor %654, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %656 = torch.aten.mul.Tensor %649, %655 : !torch.tensor, !torch.tensor -> !torch.tensor
    %657 = torch.prim.ListConstruct %int-1, %int3072 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %658 = torch.aten.view %656, %657 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %659 = torch.aten.addmm %arg108, %658, %arg109, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %660 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %661 = torch.aten.view %659, %660 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %662 = torch.aten.add.Tensor %642, %661, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %663 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_39, %result1_40, %result2_41 = torch.aten.native_layer_norm %662, %663, %arg115, %arg114, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %664 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %665 = torch.aten.view %result0_39, %664 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %666 = torch.aten.addmm %arg110, %665, %arg111, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %667 = torch.prim.ListConstruct %int4, %int512, %int2304 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %668 = torch.aten.view %666, %667 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %669 = torch.operator "aten.split.Tensor"(%668, %int768, %int2) : (!torch.tensor, !torch.int, !torch.int) -> !torch.list<!torch.tensor>
    %670 = torch.aten.__getitem__.t %669, %int0 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %671 = torch.aten.__getitem__.t %669, %int1 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %672 = torch.aten.__getitem__.t %669, %int2 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %673 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %674 = torch.aten.view %670, %673 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %675 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %676 = torch.aten.permute %674, %675 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %677 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %678 = torch.aten.view %671, %677 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %679 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %680 = torch.aten.permute %678, %679 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %681 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %682 = torch.aten.view %672, %681 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %683 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %684 = torch.aten.permute %682, %683 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %685 = torch.aten.transpose.int %680, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %686 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %687 = torch.aten.expand %676, %686, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %688 = torch.aten.clone %687, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %689 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %690 = torch.aten._unsafe_view %688, %689 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %691 = torch.prim.ListConstruct %int4, %int12, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %692 = torch.aten.expand %685, %691, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %693 = torch.aten.clone %692, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %694 = torch.prim.ListConstruct %int48, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %695 = torch.aten._unsafe_view %693, %694 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %696 = torch.aten.bmm %690, %695 : !torch.tensor, !torch.tensor -> !torch.tensor
    %697 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %698 = torch.aten._unsafe_view %696, %697 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %699 = torch.aten.div.Tensor %698, %5 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %700 = torch.aten.slice.Tensor %arg168, %int0, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %701 = torch.aten.slice.Tensor %700, %int1, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %702 = torch.aten.slice.Tensor %701, %int2, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %703 = torch.aten.slice.Tensor %702, %int3, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %704 = torch.operator "aten._to_copy"(%703, %int11, %none, %none, %none, %false, %none) : (!torch.tensor, !torch.int, !torch.none, !torch.none, !torch.none, !torch.bool, !torch.none) -> !torch.tensor
    %705 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %706 = torch.aten.expand %arg169, %705, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %707 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %708 = torch.aten.expand %699, %707, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %709 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %710 = torch.aten.expand %704, %709, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %711 = torch.operator "aten._s_where"(%710, %708, %706) : (!torch.tensor, !torch.tensor, !torch.tensor) -> !torch.tensor
    %712 = torch.aten._softmax %711, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %713 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %714 = torch.aten.expand %712, %713, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %715 = torch.prim.ListConstruct %int48, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %716 = torch.aten.view %714, %715 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %717 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %718 = torch.aten.expand %684, %717, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %719 = torch.aten.clone %718, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %720 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %721 = torch.aten._unsafe_view %719, %720 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %722 = torch.aten.bmm %716, %721 : !torch.tensor, !torch.tensor -> !torch.tensor
    %723 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %724 = torch.aten._unsafe_view %722, %723 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %725 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %726 = torch.aten.permute %724, %725 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %727 = torch.aten.clone %726, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %728 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %729 = torch.aten.view %727, %728 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %730 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %731 = torch.aten.view %729, %730 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %732 = torch.aten.addmm %arg112, %731, %arg113, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %733 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %734 = torch.aten.view %732, %733 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %735 = torch.aten.add.Tensor %734, %662, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %736 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_42, %result1_43, %result2_44 = torch.aten.native_layer_norm %735, %736, %arg117, %arg116, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %737 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %738 = torch.aten.view %result0_42, %737 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %739 = torch.aten.addmm %arg118, %738, %arg119, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %740 = torch.prim.ListConstruct %int4, %int512, %int3072 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %741 = torch.aten.view %739, %740 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %742 = torch.aten.mul.Tensor %741, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %743 = torch.aten.pow.Tensor_Scalar %741, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %744 = torch.aten.mul.Tensor %743, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %745 = torch.aten.add.Tensor %741, %744, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %746 = torch.aten.mul.Tensor %745, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %747 = torch.aten.tanh %746 : !torch.tensor -> !torch.tensor
    %748 = torch.aten.add.Tensor %747, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %749 = torch.aten.mul.Tensor %742, %748 : !torch.tensor, !torch.tensor -> !torch.tensor
    %750 = torch.prim.ListConstruct %int-1, %int3072 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %751 = torch.aten.view %749, %750 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %752 = torch.aten.addmm %arg120, %751, %arg121, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %753 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %754 = torch.aten.view %752, %753 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %755 = torch.aten.add.Tensor %735, %754, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %756 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_45, %result1_46, %result2_47 = torch.aten.native_layer_norm %755, %756, %arg127, %arg126, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %757 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %758 = torch.aten.view %result0_45, %757 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %759 = torch.aten.addmm %arg122, %758, %arg123, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %760 = torch.prim.ListConstruct %int4, %int512, %int2304 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %761 = torch.aten.view %759, %760 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %762 = torch.operator "aten.split.Tensor"(%761, %int768, %int2) : (!torch.tensor, !torch.int, !torch.int) -> !torch.list<!torch.tensor>
    %763 = torch.aten.__getitem__.t %762, %int0 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %764 = torch.aten.__getitem__.t %762, %int1 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %765 = torch.aten.__getitem__.t %762, %int2 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %766 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %767 = torch.aten.view %763, %766 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %768 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %769 = torch.aten.permute %767, %768 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %770 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %771 = torch.aten.view %764, %770 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %772 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %773 = torch.aten.permute %771, %772 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %774 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %775 = torch.aten.view %765, %774 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %776 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %777 = torch.aten.permute %775, %776 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %778 = torch.aten.transpose.int %773, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %779 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %780 = torch.aten.expand %769, %779, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %781 = torch.aten.clone %780, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %782 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %783 = torch.aten._unsafe_view %781, %782 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %784 = torch.prim.ListConstruct %int4, %int12, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %785 = torch.aten.expand %778, %784, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %786 = torch.aten.clone %785, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %787 = torch.prim.ListConstruct %int48, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %788 = torch.aten._unsafe_view %786, %787 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %789 = torch.aten.bmm %783, %788 : !torch.tensor, !torch.tensor -> !torch.tensor
    %790 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %791 = torch.aten._unsafe_view %789, %790 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %792 = torch.aten.div.Tensor %791, %5 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %793 = torch.aten.slice.Tensor %arg170, %int0, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %794 = torch.aten.slice.Tensor %793, %int1, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %795 = torch.aten.slice.Tensor %794, %int2, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %796 = torch.aten.slice.Tensor %795, %int3, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %797 = torch.operator "aten._to_copy"(%796, %int11, %none, %none, %none, %false, %none) : (!torch.tensor, !torch.int, !torch.none, !torch.none, !torch.none, !torch.bool, !torch.none) -> !torch.tensor
    %798 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %799 = torch.aten.expand %arg171, %798, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %800 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %801 = torch.aten.expand %792, %800, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %802 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %803 = torch.aten.expand %797, %802, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %804 = torch.operator "aten._s_where"(%803, %801, %799) : (!torch.tensor, !torch.tensor, !torch.tensor) -> !torch.tensor
    %805 = torch.aten._softmax %804, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %806 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %807 = torch.aten.expand %805, %806, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %808 = torch.prim.ListConstruct %int48, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %809 = torch.aten.view %807, %808 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %810 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %811 = torch.aten.expand %777, %810, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %812 = torch.aten.clone %811, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %813 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %814 = torch.aten._unsafe_view %812, %813 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %815 = torch.aten.bmm %809, %814 : !torch.tensor, !torch.tensor -> !torch.tensor
    %816 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %817 = torch.aten._unsafe_view %815, %816 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %818 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %819 = torch.aten.permute %817, %818 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %820 = torch.aten.clone %819, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %821 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %822 = torch.aten.view %820, %821 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %823 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %824 = torch.aten.view %822, %823 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %825 = torch.aten.addmm %arg124, %824, %arg125, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %826 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %827 = torch.aten.view %825, %826 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %828 = torch.aten.add.Tensor %827, %755, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %829 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_48, %result1_49, %result2_50 = torch.aten.native_layer_norm %828, %829, %arg129, %arg128, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %830 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %831 = torch.aten.view %result0_48, %830 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %832 = torch.aten.addmm %arg130, %831, %arg131, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %833 = torch.prim.ListConstruct %int4, %int512, %int3072 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %834 = torch.aten.view %832, %833 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %835 = torch.aten.mul.Tensor %834, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %836 = torch.aten.pow.Tensor_Scalar %834, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %837 = torch.aten.mul.Tensor %836, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %838 = torch.aten.add.Tensor %834, %837, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %839 = torch.aten.mul.Tensor %838, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %840 = torch.aten.tanh %839 : !torch.tensor -> !torch.tensor
    %841 = torch.aten.add.Tensor %840, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %842 = torch.aten.mul.Tensor %835, %841 : !torch.tensor, !torch.tensor -> !torch.tensor
    %843 = torch.prim.ListConstruct %int-1, %int3072 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %844 = torch.aten.view %842, %843 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %845 = torch.aten.addmm %arg132, %844, %arg133, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %846 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %847 = torch.aten.view %845, %846 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %848 = torch.aten.add.Tensor %828, %847, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %849 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_51, %result1_52, %result2_53 = torch.aten.native_layer_norm %848, %849, %arg139, %arg138, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %850 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %851 = torch.aten.view %result0_51, %850 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %852 = torch.aten.addmm %arg134, %851, %arg135, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %853 = torch.prim.ListConstruct %int4, %int512, %int2304 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %854 = torch.aten.view %852, %853 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %855 = torch.operator "aten.split.Tensor"(%854, %int768, %int2) : (!torch.tensor, !torch.int, !torch.int) -> !torch.list<!torch.tensor>
    %856 = torch.aten.__getitem__.t %855, %int0 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %857 = torch.aten.__getitem__.t %855, %int1 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %858 = torch.aten.__getitem__.t %855, %int2 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %859 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %860 = torch.aten.view %856, %859 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %861 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %862 = torch.aten.permute %860, %861 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %863 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %864 = torch.aten.view %857, %863 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %865 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %866 = torch.aten.permute %864, %865 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %867 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %868 = torch.aten.view %858, %867 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %869 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %870 = torch.aten.permute %868, %869 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %871 = torch.aten.transpose.int %866, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %872 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %873 = torch.aten.expand %862, %872, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %874 = torch.aten.clone %873, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %875 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %876 = torch.aten._unsafe_view %874, %875 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %877 = torch.prim.ListConstruct %int4, %int12, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %878 = torch.aten.expand %871, %877, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %879 = torch.aten.clone %878, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %880 = torch.prim.ListConstruct %int48, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %881 = torch.aten._unsafe_view %879, %880 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %882 = torch.aten.bmm %876, %881 : !torch.tensor, !torch.tensor -> !torch.tensor
    %883 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %884 = torch.aten._unsafe_view %882, %883 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %885 = torch.aten.div.Tensor %884, %5 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %886 = torch.aten.slice.Tensor %arg172, %int0, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %887 = torch.aten.slice.Tensor %886, %int1, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %888 = torch.aten.slice.Tensor %887, %int2, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %889 = torch.aten.slice.Tensor %888, %int3, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %890 = torch.operator "aten._to_copy"(%889, %int11, %none, %none, %none, %false, %none) : (!torch.tensor, !torch.int, !torch.none, !torch.none, !torch.none, !torch.bool, !torch.none) -> !torch.tensor
    %891 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %892 = torch.aten.expand %arg173, %891, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %893 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %894 = torch.aten.expand %885, %893, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %895 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %896 = torch.aten.expand %890, %895, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %897 = torch.operator "aten._s_where"(%896, %894, %892) : (!torch.tensor, !torch.tensor, !torch.tensor) -> !torch.tensor
    %898 = torch.aten._softmax %897, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %899 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %900 = torch.aten.expand %898, %899, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %901 = torch.prim.ListConstruct %int48, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %902 = torch.aten.view %900, %901 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %903 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %904 = torch.aten.expand %870, %903, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %905 = torch.aten.clone %904, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %906 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %907 = torch.aten._unsafe_view %905, %906 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %908 = torch.aten.bmm %902, %907 : !torch.tensor, !torch.tensor -> !torch.tensor
    %909 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %910 = torch.aten._unsafe_view %908, %909 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %911 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %912 = torch.aten.permute %910, %911 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %913 = torch.aten.clone %912, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %914 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %915 = torch.aten.view %913, %914 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %916 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %917 = torch.aten.view %915, %916 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %918 = torch.aten.addmm %arg136, %917, %arg137, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %919 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %920 = torch.aten.view %918, %919 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %921 = torch.aten.add.Tensor %920, %848, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %922 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_54, %result1_55, %result2_56 = torch.aten.native_layer_norm %921, %922, %arg141, %arg140, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %923 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %924 = torch.aten.view %result0_54, %923 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %925 = torch.aten.addmm %arg142, %924, %arg143, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %926 = torch.prim.ListConstruct %int4, %int512, %int3072 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %927 = torch.aten.view %925, %926 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %928 = torch.aten.mul.Tensor %927, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %929 = torch.aten.pow.Tensor_Scalar %927, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %930 = torch.aten.mul.Tensor %929, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %931 = torch.aten.add.Tensor %927, %930, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %932 = torch.aten.mul.Tensor %931, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %933 = torch.aten.tanh %932 : !torch.tensor -> !torch.tensor
    %934 = torch.aten.add.Tensor %933, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %935 = torch.aten.mul.Tensor %928, %934 : !torch.tensor, !torch.tensor -> !torch.tensor
    %936 = torch.prim.ListConstruct %int-1, %int3072 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %937 = torch.aten.view %935, %936 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %938 = torch.aten.addmm %arg144, %937, %arg145, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %939 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %940 = torch.aten.view %938, %939 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %941 = torch.aten.add.Tensor %921, %940, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %942 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_57, %result1_58, %result2_59 = torch.aten.native_layer_norm %941, %942, %arg31, %arg30, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %943 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %944 = torch.aten.view %result0_57, %943 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %945 = torch.aten.addmm %arg26, %944, %arg27, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %946 = torch.prim.ListConstruct %int4, %int512, %int2304 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %947 = torch.aten.view %945, %946 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %948 = torch.operator "aten.split.Tensor"(%947, %int768, %int2) : (!torch.tensor, !torch.int, !torch.int) -> !torch.list<!torch.tensor>
    %949 = torch.aten.__getitem__.t %948, %int0 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %950 = torch.aten.__getitem__.t %948, %int1 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %951 = torch.aten.__getitem__.t %948, %int2 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %952 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %953 = torch.aten.view %949, %952 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %954 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %955 = torch.aten.permute %953, %954 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %956 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %957 = torch.aten.view %950, %956 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %958 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %959 = torch.aten.permute %957, %958 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %960 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %961 = torch.aten.view %951, %960 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %962 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %963 = torch.aten.permute %961, %962 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %964 = torch.aten.transpose.int %959, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %965 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %966 = torch.aten.expand %955, %965, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %967 = torch.aten.clone %966, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %968 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %969 = torch.aten._unsafe_view %967, %968 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %970 = torch.prim.ListConstruct %int4, %int12, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %971 = torch.aten.expand %964, %970, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %972 = torch.aten.clone %971, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %973 = torch.prim.ListConstruct %int48, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %974 = torch.aten._unsafe_view %972, %973 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %975 = torch.aten.bmm %969, %974 : !torch.tensor, !torch.tensor -> !torch.tensor
    %976 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %977 = torch.aten._unsafe_view %975, %976 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %978 = torch.aten.div.Tensor %977, %5 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %979 = torch.aten.slice.Tensor %arg154, %int0, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %980 = torch.aten.slice.Tensor %979, %int1, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %981 = torch.aten.slice.Tensor %980, %int2, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %982 = torch.aten.slice.Tensor %981, %int3, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %983 = torch.operator "aten._to_copy"(%982, %int11, %none, %none, %none, %false, %none) : (!torch.tensor, !torch.int, !torch.none, !torch.none, !torch.none, !torch.bool, !torch.none) -> !torch.tensor
    %984 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %985 = torch.aten.expand %arg155, %984, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %986 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %987 = torch.aten.expand %978, %986, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %988 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %989 = torch.aten.expand %983, %988, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %990 = torch.operator "aten._s_where"(%989, %987, %985) : (!torch.tensor, !torch.tensor, !torch.tensor) -> !torch.tensor
    %991 = torch.aten._softmax %990, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %992 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %993 = torch.aten.expand %991, %992, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %994 = torch.prim.ListConstruct %int48, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %995 = torch.aten.view %993, %994 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %996 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %997 = torch.aten.expand %963, %996, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %998 = torch.aten.clone %997, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %999 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1000 = torch.aten._unsafe_view %998, %999 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1001 = torch.aten.bmm %995, %1000 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1002 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1003 = torch.aten._unsafe_view %1001, %1002 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1004 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1005 = torch.aten.permute %1003, %1004 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1006 = torch.aten.clone %1005, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1007 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1008 = torch.aten.view %1006, %1007 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1009 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1010 = torch.aten.view %1008, %1009 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1011 = torch.aten.addmm %arg28, %1010, %arg29, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1012 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1013 = torch.aten.view %1011, %1012 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1014 = torch.aten.add.Tensor %1013, %941, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1015 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_60, %result1_61, %result2_62 = torch.aten.native_layer_norm %1014, %1015, %arg33, %arg32, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1016 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1017 = torch.aten.view %result0_60, %1016 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1018 = torch.aten.addmm %arg34, %1017, %arg35, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1019 = torch.prim.ListConstruct %int4, %int512, %int3072 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1020 = torch.aten.view %1018, %1019 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1021 = torch.aten.mul.Tensor %1020, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1022 = torch.aten.pow.Tensor_Scalar %1020, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %1023 = torch.aten.mul.Tensor %1022, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1024 = torch.aten.add.Tensor %1020, %1023, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1025 = torch.aten.mul.Tensor %1024, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1026 = torch.aten.tanh %1025 : !torch.tensor -> !torch.tensor
    %1027 = torch.aten.add.Tensor %1026, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %1028 = torch.aten.mul.Tensor %1021, %1027 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1029 = torch.prim.ListConstruct %int-1, %int3072 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1030 = torch.aten.view %1028, %1029 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1031 = torch.aten.addmm %arg36, %1030, %arg37, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1032 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1033 = torch.aten.view %1031, %1032 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1034 = torch.aten.add.Tensor %1014, %1033, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1035 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_63, %result1_64, %result2_65 = torch.aten.native_layer_norm %1034, %1035, %arg43, %arg42, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1036 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1037 = torch.aten.view %result0_63, %1036 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1038 = torch.aten.addmm %arg38, %1037, %arg39, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1039 = torch.prim.ListConstruct %int4, %int512, %int2304 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1040 = torch.aten.view %1038, %1039 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1041 = torch.operator "aten.split.Tensor"(%1040, %int768, %int2) : (!torch.tensor, !torch.int, !torch.int) -> !torch.list<!torch.tensor>
    %1042 = torch.aten.__getitem__.t %1041, %int0 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %1043 = torch.aten.__getitem__.t %1041, %int1 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %1044 = torch.aten.__getitem__.t %1041, %int2 : !torch.list<!torch.tensor>, !torch.int -> !torch.tensor
    %1045 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1046 = torch.aten.view %1042, %1045 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1047 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1048 = torch.aten.permute %1046, %1047 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1049 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1050 = torch.aten.view %1043, %1049 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1051 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1052 = torch.aten.permute %1050, %1051 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1053 = torch.prim.ListConstruct %int4, %int512, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1054 = torch.aten.view %1044, %1053 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1055 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1056 = torch.aten.permute %1054, %1055 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1057 = torch.aten.transpose.int %1052, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1058 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1059 = torch.aten.expand %1048, %1058, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1060 = torch.aten.clone %1059, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1061 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1062 = torch.aten._unsafe_view %1060, %1061 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1063 = torch.prim.ListConstruct %int4, %int12, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1064 = torch.aten.expand %1057, %1063, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1065 = torch.aten.clone %1064, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1066 = torch.prim.ListConstruct %int48, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1067 = torch.aten._unsafe_view %1065, %1066 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1068 = torch.aten.bmm %1062, %1067 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1069 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1070 = torch.aten._unsafe_view %1068, %1069 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1071 = torch.aten.div.Tensor %1070, %5 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1072 = torch.aten.slice.Tensor %arg156, %int0, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %1073 = torch.aten.slice.Tensor %1072, %int1, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %1074 = torch.aten.slice.Tensor %1073, %int2, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %1075 = torch.aten.slice.Tensor %1074, %int3, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %1076 = torch.operator "aten._to_copy"(%1075, %int11, %none, %none, %none, %false, %none) : (!torch.tensor, !torch.int, !torch.none, !torch.none, !torch.none, !torch.bool, !torch.none) -> !torch.tensor
    %1077 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1078 = torch.aten.expand %arg157, %1077, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1079 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1080 = torch.aten.expand %1071, %1079, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1081 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1082 = torch.aten.expand %1076, %1081, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1083 = torch.operator "aten._s_where"(%1082, %1080, %1078) : (!torch.tensor, !torch.tensor, !torch.tensor) -> !torch.tensor
    %1084 = torch.aten._softmax %1083, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %1085 = torch.prim.ListConstruct %int4, %int12, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1086 = torch.aten.expand %1084, %1085, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1087 = torch.prim.ListConstruct %int48, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1088 = torch.aten.view %1086, %1087 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1089 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1090 = torch.aten.expand %1056, %1089, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1091 = torch.aten.clone %1090, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1092 = torch.prim.ListConstruct %int48, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1093 = torch.aten._unsafe_view %1091, %1092 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1094 = torch.aten.bmm %1088, %1093 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1095 = torch.prim.ListConstruct %int4, %int12, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1096 = torch.aten._unsafe_view %1094, %1095 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1097 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1098 = torch.aten.permute %1096, %1097 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1099 = torch.aten.clone %1098, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1100 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1101 = torch.aten.view %1099, %1100 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1102 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1103 = torch.aten.view %1101, %1102 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1104 = torch.aten.addmm %arg40, %1103, %arg41, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1105 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1106 = torch.aten.view %1104, %1105 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1107 = torch.aten.add.Tensor %1106, %1034, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1108 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_66, %result1_67, %result2_68 = torch.aten.native_layer_norm %1107, %1108, %arg45, %arg44, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1109 = torch.prim.ListConstruct %int-1, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1110 = torch.aten.view %result0_66, %1109 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1111 = torch.aten.addmm %arg46, %1110, %arg47, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1112 = torch.prim.ListConstruct %int4, %int512, %int3072 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1113 = torch.aten.view %1111, %1112 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1114 = torch.aten.mul.Tensor %1113, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1115 = torch.aten.pow.Tensor_Scalar %1113, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %1116 = torch.aten.mul.Tensor %1115, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1117 = torch.aten.add.Tensor %1113, %1116, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1118 = torch.aten.mul.Tensor %1117, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1119 = torch.aten.tanh %1118 : !torch.tensor -> !torch.tensor
    %1120 = torch.aten.add.Tensor %1119, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %1121 = torch.aten.mul.Tensor %1114, %1120 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1122 = torch.prim.ListConstruct %int-1, %int3072 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1123 = torch.aten.view %1121, %1122 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1124 = torch.aten.addmm %arg48, %1123, %arg49, %int1, %int1 : !torch.tensor, !torch.tensor, !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1125 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1126 = torch.aten.view %1124, %1125 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1127 = torch.aten.add.Tensor %1107, %1126, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1128 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<!torch.int>
    %result0_69, %result1_70, %result2_71 = torch.aten.native_layer_norm %1127, %1128, %arg147, %arg146, %float1.000000e-05 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1129 = torch.prim.ListConstruct %int4, %int512, %int768 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1130 = torch.aten.view %result0_69, %1129 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1131 = torch.aten.t %arg1 : !torch.tensor -> !torch.tensor
    %1132 = torch.prim.ListConstruct %int2048, %int768 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1133 = torch.aten.view %1130, %1132 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1134 = torch.aten.mm %1133, %1131 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1135 = torch.prim.ListConstruct %int4, %int512, %int50257 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1136 = torch.aten._unsafe_view %1134, %1135 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1137 = torch.aten.slice.Tensor %1136, %int1, %int0, %int-1, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %1138 = torch.aten.slice.Tensor %1137, %int2, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %1139 = torch.aten.clone %1138, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1140 = torch.aten.slice.Tensor %arg175, %int1, %int1, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %1141 = torch.aten.clone %1140, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1142 = torch.prim.ListConstruct %int-1, %int50257 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1143 = torch.aten.view %1139, %1142 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1144 = torch.prim.ListConstruct %int-1 : (!torch.int) -> !torch.list<!torch.int>
    %1145 = torch.aten.view %1141, %1144 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1146 = torch.aten._log_softmax %1143, %int1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %output, %total_weight = torch.aten.nll_loss_forward %1146, %1145, %none, %int1, %int-100 : !torch.tensor, !torch.tensor, !torch.none, !torch.int, !torch.int -> !torch.tensor, !torch.tensor
    %1147 = torch.prim.ListConstruct %output, %1136 : (!torch.tensor, !torch.tensor) -> !torch.list<!torch.tensor>
    return %1147 : !torch.list<!torch.tensor>
  }
  torch.class_type @__torch__.torch.fx.graph_module.___torch_mangle_0.GraphModule {
    torch.method "forward", @__torch__.torch.fx.graph_module.___torch_mangle_0.GraphModule.forward
  }
  %0 = torch.nn_module {
  } : !torch.nn.Module<"__torch__.torch.fx.graph_module.___torch_mangle_0.GraphModule">
}