dump1.mlir

module attributes {torch.debug_module_name = "_lambda"} {
  func.func private @__torch__.torch.fx.graph_module._lambda.forward(%arg0: !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda">, %arg1: !torch.tensor {torch.type_bound = !torch.vtensor<[1,77],si64>}) -> !torch.tensor {
    %199 = torch.tensor_static_info_cast %arg1 : !torch.tensor to !torch.tensor<[1,77],si64>
    %200 = torch.prim.GetAttr %arg0["_param_constant195"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %201 = torch.prim.GetAttr %arg0["_param_constant194"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %202 = torch.prim.GetAttr %arg0["_param_constant193"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %203 = torch.prim.GetAttr %arg0["_param_constant192"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %204 = torch.prim.GetAttr %arg0["_param_constant191"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %205 = torch.prim.GetAttr %arg0["_param_constant190"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %206 = torch.prim.GetAttr %arg0["_param_constant189"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %207 = torch.prim.GetAttr %arg0["_param_constant188"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %208 = torch.prim.GetAttr %arg0["_param_constant187"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %209 = torch.prim.GetAttr %arg0["_param_constant186"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %210 = torch.prim.GetAttr %arg0["_param_constant185"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %211 = torch.prim.GetAttr %arg0["_param_constant184"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %212 = torch.prim.GetAttr %arg0["_param_constant183"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %213 = torch.prim.GetAttr %arg0["_param_constant182"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %214 = torch.prim.GetAttr %arg0["_param_constant181"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %215 = torch.prim.GetAttr %arg0["_param_constant180"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %216 = torch.prim.GetAttr %arg0["_param_constant179"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %217 = torch.prim.GetAttr %arg0["_param_constant178"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %218 = torch.prim.GetAttr %arg0["_param_constant177"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %219 = torch.prim.GetAttr %arg0["_param_constant176"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %220 = torch.prim.GetAttr %arg0["_param_constant175"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %221 = torch.prim.GetAttr %arg0["_param_constant174"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %222 = torch.prim.GetAttr %arg0["_param_constant173"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %223 = torch.prim.GetAttr %arg0["_param_constant172"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %224 = torch.prim.GetAttr %arg0["_param_constant171"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %225 = torch.prim.GetAttr %arg0["_param_constant170"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %226 = torch.prim.GetAttr %arg0["_param_constant169"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %227 = torch.prim.GetAttr %arg0["_param_constant168"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %228 = torch.prim.GetAttr %arg0["_param_constant167"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %229 = torch.prim.GetAttr %arg0["_param_constant166"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %230 = torch.prim.GetAttr %arg0["_param_constant165"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %231 = torch.prim.GetAttr %arg0["_param_constant164"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %232 = torch.prim.GetAttr %arg0["_param_constant163"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %233 = torch.prim.GetAttr %arg0["_param_constant162"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %234 = torch.prim.GetAttr %arg0["_param_constant161"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %235 = torch.prim.GetAttr %arg0["_param_constant160"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %236 = torch.prim.GetAttr %arg0["_param_constant159"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %237 = torch.prim.GetAttr %arg0["_param_constant158"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %238 = torch.prim.GetAttr %arg0["_param_constant157"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %239 = torch.prim.GetAttr %arg0["_param_constant156"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %240 = torch.prim.GetAttr %arg0["_param_constant155"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %241 = torch.prim.GetAttr %arg0["_param_constant154"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %242 = torch.prim.GetAttr %arg0["_param_constant153"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %243 = torch.prim.GetAttr %arg0["_param_constant152"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %244 = torch.prim.GetAttr %arg0["_param_constant151"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %245 = torch.prim.GetAttr %arg0["_param_constant150"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %246 = torch.prim.GetAttr %arg0["_param_constant149"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %247 = torch.prim.GetAttr %arg0["_param_constant148"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %248 = torch.prim.GetAttr %arg0["_param_constant147"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %249 = torch.prim.GetAttr %arg0["_param_constant146"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %250 = torch.prim.GetAttr %arg0["_param_constant145"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %251 = torch.prim.GetAttr %arg0["_param_constant144"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %252 = torch.prim.GetAttr %arg0["_param_constant143"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %253 = torch.prim.GetAttr %arg0["_param_constant142"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %254 = torch.prim.GetAttr %arg0["_param_constant141"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %255 = torch.prim.GetAttr %arg0["_param_constant140"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %256 = torch.prim.GetAttr %arg0["_param_constant139"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %257 = torch.prim.GetAttr %arg0["_param_constant138"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %258 = torch.prim.GetAttr %arg0["_param_constant137"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %259 = torch.prim.GetAttr %arg0["_param_constant136"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %260 = torch.prim.GetAttr %arg0["_param_constant135"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %261 = torch.prim.GetAttr %arg0["_param_constant134"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %262 = torch.prim.GetAttr %arg0["_param_constant133"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %263 = torch.prim.GetAttr %arg0["_param_constant132"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %264 = torch.prim.GetAttr %arg0["_param_constant131"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %265 = torch.prim.GetAttr %arg0["_param_constant130"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %266 = torch.prim.GetAttr %arg0["_param_constant129"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %267 = torch.prim.GetAttr %arg0["_param_constant128"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %268 = torch.prim.GetAttr %arg0["_param_constant127"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %269 = torch.prim.GetAttr %arg0["_param_constant126"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %270 = torch.prim.GetAttr %arg0["_param_constant125"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %271 = torch.prim.GetAttr %arg0["_param_constant124"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %272 = torch.prim.GetAttr %arg0["_param_constant123"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %273 = torch.prim.GetAttr %arg0["_param_constant122"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %274 = torch.prim.GetAttr %arg0["_param_constant121"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %275 = torch.prim.GetAttr %arg0["_param_constant120"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %276 = torch.prim.GetAttr %arg0["_param_constant119"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %277 = torch.prim.GetAttr %arg0["_param_constant118"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %278 = torch.prim.GetAttr %arg0["_param_constant117"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %279 = torch.prim.GetAttr %arg0["_param_constant116"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %280 = torch.prim.GetAttr %arg0["_param_constant115"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %281 = torch.prim.GetAttr %arg0["_param_constant114"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %282 = torch.prim.GetAttr %arg0["_param_constant113"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %283 = torch.prim.GetAttr %arg0["_param_constant112"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %284 = torch.prim.GetAttr %arg0["_param_constant111"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %285 = torch.prim.GetAttr %arg0["_param_constant110"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %286 = torch.prim.GetAttr %arg0["_param_constant109"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %287 = torch.prim.GetAttr %arg0["_param_constant108"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %288 = torch.prim.GetAttr %arg0["_param_constant107"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %289 = torch.prim.GetAttr %arg0["_param_constant106"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %290 = torch.prim.GetAttr %arg0["_param_constant105"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %291 = torch.prim.GetAttr %arg0["_param_constant104"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %292 = torch.prim.GetAttr %arg0["_param_constant103"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %293 = torch.prim.GetAttr %arg0["_param_constant102"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %294 = torch.prim.GetAttr %arg0["_param_constant101"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %295 = torch.prim.GetAttr %arg0["_param_constant100"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %296 = torch.prim.GetAttr %arg0["_param_constant99"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %297 = torch.prim.GetAttr %arg0["_param_constant98"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %298 = torch.prim.GetAttr %arg0["_param_constant97"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %299 = torch.prim.GetAttr %arg0["_param_constant96"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %300 = torch.prim.GetAttr %arg0["_param_constant95"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %301 = torch.prim.GetAttr %arg0["_param_constant94"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %302 = torch.prim.GetAttr %arg0["_param_constant93"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %303 = torch.prim.GetAttr %arg0["_param_constant92"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %304 = torch.prim.GetAttr %arg0["_param_constant91"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %305 = torch.prim.GetAttr %arg0["_param_constant90"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %306 = torch.prim.GetAttr %arg0["_param_constant89"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %307 = torch.prim.GetAttr %arg0["_param_constant88"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %308 = torch.prim.GetAttr %arg0["_param_constant87"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %309 = torch.prim.GetAttr %arg0["_param_constant86"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %310 = torch.prim.GetAttr %arg0["_param_constant85"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %311 = torch.prim.GetAttr %arg0["_param_constant84"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %312 = torch.prim.GetAttr %arg0["_param_constant83"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %313 = torch.prim.GetAttr %arg0["_param_constant82"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %314 = torch.prim.GetAttr %arg0["_param_constant81"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %315 = torch.prim.GetAttr %arg0["_param_constant80"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %316 = torch.prim.GetAttr %arg0["_param_constant79"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %317 = torch.prim.GetAttr %arg0["_param_constant78"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %318 = torch.prim.GetAttr %arg0["_param_constant77"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %319 = torch.prim.GetAttr %arg0["_param_constant76"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %320 = torch.prim.GetAttr %arg0["_param_constant75"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %321 = torch.prim.GetAttr %arg0["_param_constant74"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %322 = torch.prim.GetAttr %arg0["_param_constant73"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %323 = torch.prim.GetAttr %arg0["_param_constant72"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %324 = torch.prim.GetAttr %arg0["_param_constant71"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %325 = torch.prim.GetAttr %arg0["_param_constant70"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %326 = torch.prim.GetAttr %arg0["_param_constant69"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %327 = torch.prim.GetAttr %arg0["_param_constant68"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %328 = torch.prim.GetAttr %arg0["_param_constant67"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %329 = torch.prim.GetAttr %arg0["_param_constant66"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %330 = torch.prim.GetAttr %arg0["_param_constant65"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %331 = torch.prim.GetAttr %arg0["_param_constant64"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %332 = torch.prim.GetAttr %arg0["_param_constant63"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %333 = torch.prim.GetAttr %arg0["_param_constant62"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %334 = torch.prim.GetAttr %arg0["_param_constant61"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %335 = torch.prim.GetAttr %arg0["_param_constant60"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %336 = torch.prim.GetAttr %arg0["_param_constant59"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %337 = torch.prim.GetAttr %arg0["_param_constant58"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %338 = torch.prim.GetAttr %arg0["_param_constant57"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %339 = torch.prim.GetAttr %arg0["_param_constant56"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %340 = torch.prim.GetAttr %arg0["_param_constant55"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %341 = torch.prim.GetAttr %arg0["_param_constant54"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %342 = torch.prim.GetAttr %arg0["_param_constant53"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %343 = torch.prim.GetAttr %arg0["_param_constant52"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %344 = torch.prim.GetAttr %arg0["_param_constant51"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %345 = torch.prim.GetAttr %arg0["_param_constant50"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %346 = torch.prim.GetAttr %arg0["_param_constant49"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %347 = torch.prim.GetAttr %arg0["_param_constant48"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %348 = torch.prim.GetAttr %arg0["_param_constant47"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %349 = torch.prim.GetAttr %arg0["_param_constant46"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %350 = torch.prim.GetAttr %arg0["_param_constant45"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %351 = torch.prim.GetAttr %arg0["_param_constant44"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %352 = torch.prim.GetAttr %arg0["_param_constant43"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %353 = torch.prim.GetAttr %arg0["_param_constant42"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %354 = torch.prim.GetAttr %arg0["_param_constant41"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %355 = torch.prim.GetAttr %arg0["_param_constant40"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %356 = torch.prim.GetAttr %arg0["_param_constant39"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %357 = torch.prim.GetAttr %arg0["_param_constant38"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %358 = torch.prim.GetAttr %arg0["_param_constant37"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %359 = torch.prim.GetAttr %arg0["_param_constant36"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %360 = torch.prim.GetAttr %arg0["_param_constant35"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %361 = torch.prim.GetAttr %arg0["_param_constant34"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %362 = torch.prim.GetAttr %arg0["_param_constant33"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %363 = torch.prim.GetAttr %arg0["_param_constant32"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %364 = torch.prim.GetAttr %arg0["_param_constant31"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %365 = torch.prim.GetAttr %arg0["_param_constant30"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %366 = torch.prim.GetAttr %arg0["_param_constant29"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %367 = torch.prim.GetAttr %arg0["_param_constant28"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %368 = torch.prim.GetAttr %arg0["_param_constant27"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %369 = torch.prim.GetAttr %arg0["_param_constant26"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %370 = torch.prim.GetAttr %arg0["_param_constant25"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %371 = torch.prim.GetAttr %arg0["_param_constant24"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %372 = torch.prim.GetAttr %arg0["_param_constant23"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %373 = torch.prim.GetAttr %arg0["_param_constant22"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %374 = torch.prim.GetAttr %arg0["_param_constant21"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %375 = torch.prim.GetAttr %arg0["_param_constant20"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %376 = torch.prim.GetAttr %arg0["_param_constant19"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %377 = torch.prim.GetAttr %arg0["_param_constant18"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %378 = torch.prim.GetAttr %arg0["_param_constant17"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %379 = torch.prim.GetAttr %arg0["_param_constant16"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %380 = torch.prim.GetAttr %arg0["_param_constant15"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %381 = torch.prim.GetAttr %arg0["_param_constant14"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %382 = torch.prim.GetAttr %arg0["_param_constant13"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %383 = torch.prim.GetAttr %arg0["_param_constant12"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %384 = torch.prim.GetAttr %arg0["_param_constant11"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %385 = torch.prim.GetAttr %arg0["_param_constant10"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %386 = torch.prim.GetAttr %arg0["_param_constant9"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %387 = torch.prim.GetAttr %arg0["_param_constant8"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %388 = torch.prim.GetAttr %arg0["_param_constant7"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %389 = torch.prim.GetAttr %arg0["_param_constant6"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %390 = torch.prim.GetAttr %arg0["_param_constant5"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %391 = torch.prim.GetAttr %arg0["_param_constant4"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %392 = torch.prim.GetAttr %arg0["_param_constant3"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %393 = torch.prim.GetAttr %arg0["_param_constant2"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %394 = torch.prim.GetAttr %arg0["_tensor_constant1"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %395 = torch.prim.GetAttr %arg0["_param_constant1"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %396 = torch.prim.GetAttr %arg0["_param_constant0"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %397 = torch.prim.GetAttr %arg0["_tensor_constant0"] : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda"> -> !torch.tensor
    %int-1 = torch.constant.int -1
    %int77 = torch.constant.int 77
    %398 = torch.prim.ListConstruct %int-1, %int77 : (!torch.int, !torch.int) -> !torch.list<int>
    %399 = torch.aten.view %199, %398 : !torch.tensor<[1,77],si64>, !torch.list<int> -> !torch.tensor<[1,77],si64>
    %int0 = torch.constant.int 0
    %int0_0 = torch.constant.int 0
    %int9223372036854775807 = torch.constant.int 9223372036854775807
    %int1 = torch.constant.int 1
    %400 = torch.aten.slice.Tensor %397, %int0, %int0_0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor<[1,77],si64>
    %int-1_1 = torch.constant.int -1
    %false = torch.constant.bool false
    %false_2 = torch.constant.bool false
    %401 = torch.aten.embedding %396, %399, %int-1_1, %false, %false_2 : !torch.tensor, !torch.tensor<[1,77],si64>, !torch.int, !torch.bool, !torch.bool -> !torch.tensor<[1,77,768],f16>
    %int-1_3 = torch.constant.int -1
    %false_4 = torch.constant.bool false
    %false_5 = torch.constant.bool false
    %402 = torch.aten.embedding %395, %400, %int-1_3, %false_4, %false_5 : !torch.tensor, !torch.tensor<[1,77],si64>, !torch.int, !torch.bool, !torch.bool -> !torch.tensor<[1,77,768],f16>
    %int1_6 = torch.constant.int 1
    %403 = torch.aten.add.Tensor %401, %402, %int1_6 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int1_7 = torch.constant.int 1
    %int77_8 = torch.constant.int 77
    %int77_9 = torch.constant.int 77
    %404 = torch.prim.ListConstruct %int1_7, %int77_8, %int77_9 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %int5 = torch.constant.int 5
    %none_10 = torch.constant.none
    %cpu = torch.constant.device "cpu"
    %false_11 = torch.constant.bool false
    %none_12 = torch.constant.none
    %405 = torch.aten.empty.memory_format %404, %int5, %none_10, %cpu, %false_11, %none_12 : !torch.list<int>, !torch.int, !torch.none, !torch.Device, !torch.bool, !torch.none -> !torch.tensor<[1,77,77],f16>
    %406 = torch.aten.lift_fresh_copy %394 : !torch.tensor -> !torch.tensor<[],f32>
    %407 = torch.aten.fill_.Tensor %405, %406 : !torch.tensor<[1,77,77],f16>, !torch.tensor<[],f32> -> !torch.tensor<[1,77,77],f16>
    %int1_13 = torch.constant.int 1
    %408 = torch.aten.triu_ %407, %int1_13 : !torch.tensor<[1,77,77],f16>, !torch.int -> !torch.tensor<[1,77,77],f16>
    %int1_14 = torch.constant.int 1
    %409 = torch.aten.unsqueeze %408, %int1_14 : !torch.tensor<[1,77,77],f16>, !torch.int -> !torch.tensor<[1,1,77,77],f16>
    %int5_15 = torch.constant.int 5
    %int0_16 = torch.constant.int 0
    %cuda3A0 = torch.constant.device "cuda:0"
    %none_17 = torch.constant.none
    %false_18 = torch.constant.bool false
    %none_19 = torch.constant.none
    %410 = torch.aten._to_copy %409, %int5_15, %int0_16, %cuda3A0, %none_17, %false_18, %none_19 : !torch.tensor<[1,1,77,77],f16>, !torch.int, !torch.int, !torch.Device, !torch.none, !torch.bool, !torch.none -> !torch.tensor<[1,1,77,77],f16>
    %int768 = torch.constant.int 768
    %411 = torch.prim.ListConstruct %int768 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05 = torch.constant.float 1.000000e-05
    %result0, %result1, %result2 = torch.aten.native_layer_norm %403, %411, %393, %392, %float1.000000e-05 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %412 = torch.aten.t %391 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_20 = torch.constant.int 77
    %int768_21 = torch.constant.int 768
    %413 = torch.prim.ListConstruct %int77_20, %int768_21 : (!torch.int, !torch.int) -> !torch.list<int>
    %414 = torch.aten.view %result0, %413 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_22 = torch.constant.int 1
    %int1_23 = torch.constant.int 1
    %415 = torch.aten.addmm %390, %414, %412, %int1_22, %int1_23 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_24 = torch.constant.int 1
    %int77_25 = torch.constant.int 77
    %int768_26 = torch.constant.int 768
    %416 = torch.prim.ListConstruct %int1_24, %int77_25, %int768_26 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %417 = torch.aten.view %415, %416 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %418 = torch.tensor.literal(dense<1.250000e-01> : tensor<f64>) : !torch.tensor<[],f64>
    %419 = torch.aten.mul.Tensor %417, %418 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,768],f16>
    %420 = torch.aten.t %389 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_27 = torch.constant.int 77
    %int768_28 = torch.constant.int 768
    %421 = torch.prim.ListConstruct %int77_27, %int768_28 : (!torch.int, !torch.int) -> !torch.list<int>
    %422 = torch.aten.view %result0, %421 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_29 = torch.constant.int 1
    %int1_30 = torch.constant.int 1
    %423 = torch.aten.addmm %388, %422, %420, %int1_29, %int1_30 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_31 = torch.constant.int 1
    %int77_32 = torch.constant.int 77
    %int768_33 = torch.constant.int 768
    %424 = torch.prim.ListConstruct %int1_31, %int77_32, %int768_33 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %425 = torch.aten.view %423, %424 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_34 = torch.constant.int 1
    %int-1_35 = torch.constant.int -1
    %int12 = torch.constant.int 12
    %int64 = torch.constant.int 64
    %426 = torch.prim.ListConstruct %int1_34, %int-1_35, %int12, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %427 = torch.aten.view %425, %426 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_36 = torch.constant.int 1
    %int2 = torch.constant.int 2
    %428 = torch.aten.transpose.int %427, %int1_36, %int2 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_37 = torch.constant.int 0
    %429 = torch.aten.clone %428, %int0_37 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %430 = torch.aten.t %387 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_38 = torch.constant.int 77
    %int768_39 = torch.constant.int 768
    %431 = torch.prim.ListConstruct %int77_38, %int768_39 : (!torch.int, !torch.int) -> !torch.list<int>
    %432 = torch.aten.view %result0, %431 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_40 = torch.constant.int 1
    %int1_41 = torch.constant.int 1
    %433 = torch.aten.addmm %386, %432, %430, %int1_40, %int1_41 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_42 = torch.constant.int 1
    %int77_43 = torch.constant.int 77
    %int768_44 = torch.constant.int 768
    %434 = torch.prim.ListConstruct %int1_42, %int77_43, %int768_44 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %435 = torch.aten.view %433, %434 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_45 = torch.constant.int 1
    %int-1_46 = torch.constant.int -1
    %int12_47 = torch.constant.int 12
    %int64_48 = torch.constant.int 64
    %436 = torch.prim.ListConstruct %int1_45, %int-1_46, %int12_47, %int64_48 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %437 = torch.aten.view %435, %436 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_49 = torch.constant.int 1
    %int2_50 = torch.constant.int 2
    %438 = torch.aten.transpose.int %437, %int1_49, %int2_50 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_51 = torch.constant.int 0
    %439 = torch.aten.clone %438, %int0_51 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int1_52 = torch.constant.int 1
    %int77_53 = torch.constant.int 77
    %int12_54 = torch.constant.int 12
    %int64_55 = torch.constant.int 64
    %440 = torch.prim.ListConstruct %int1_52, %int77_53, %int12_54, %int64_55 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %441 = torch.aten.view %419, %440 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_56 = torch.constant.int 1
    %int2_57 = torch.constant.int 2
    %442 = torch.aten.transpose.int %441, %int1_56, %int2_57 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_58 = torch.constant.int 0
    %443 = torch.aten.clone %442, %int0_58 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int12_59 = torch.constant.int 12
    %int-1_60 = torch.constant.int -1
    %int64_61 = torch.constant.int 64
    %444 = torch.prim.ListConstruct %int12_59, %int-1_60, %int64_61 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %445 = torch.aten.view %443, %444 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_62 = torch.constant.int 12
    %int-1_63 = torch.constant.int -1
    %int64_64 = torch.constant.int 64
    %446 = torch.prim.ListConstruct %int12_62, %int-1_63, %int64_64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %447 = torch.aten.view %429, %446 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_65 = torch.constant.int 12
    %int-1_66 = torch.constant.int -1
    %int64_67 = torch.constant.int 64
    %448 = torch.prim.ListConstruct %int12_65, %int-1_66, %int64_67 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %449 = torch.aten.view %439, %448 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int1_68 = torch.constant.int 1
    %int2_69 = torch.constant.int 2
    %450 = torch.aten.transpose.int %447, %int1_68, %int2_69 : !torch.tensor<[12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[12,64,77],f16>
    %451 = torch.aten.bmm %445, %450 : !torch.tensor<[12,77,64],f16>, !torch.tensor<[12,64,77],f16> -> !torch.tensor<[12,77,77],f16>
    %int1_70 = torch.constant.int 1
    %int12_71 = torch.constant.int 12
    %int77_72 = torch.constant.int 77
    %int77_73 = torch.constant.int 77
    %452 = torch.prim.ListConstruct %int1_70, %int12_71, %int77_72, %int77_73 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %453 = torch.aten.view %451, %452 : !torch.tensor<[12,77,77],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,77],f16>
    %int1_74 = torch.constant.int 1
    %454 = torch.aten.add.Tensor %453, %410, %int1_74 : !torch.tensor<[1,12,77,77],f16>, !torch.tensor<[1,1,77,77],f16>, !torch.int -> !torch.tensor<[1,12,77,77],f16>
    %int12_75 = torch.constant.int 12
    %int77_76 = torch.constant.int 77
    %int77_77 = torch.constant.int 77
    %455 = torch.prim.ListConstruct %int12_75, %int77_76, %int77_77 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %456 = torch.aten.view %454, %455 : !torch.tensor<[1,12,77,77],f16>, !torch.list<int> -> !torch.tensor<[12,77,77],f16>
    %int-1_78 = torch.constant.int -1
    %false_79 = torch.constant.bool false
    %457 = torch.aten._softmax %456, %int-1_78, %false_79 : !torch.tensor<[12,77,77],f16>, !torch.int, !torch.bool -> !torch.tensor<[12,77,77],f16>
    %458 = torch.aten.bmm %457, %449 : !torch.tensor<[12,77,77],f16>, !torch.tensor<[12,77,64],f16> -> !torch.tensor<[12,77,64],f16>
    %int1_80 = torch.constant.int 1
    %int12_81 = torch.constant.int 12
    %int77_82 = torch.constant.int 77
    %int64_83 = torch.constant.int 64
    %459 = torch.prim.ListConstruct %int1_80, %int12_81, %int77_82, %int64_83 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %460 = torch.aten.view %458, %459 : !torch.tensor<[12,77,64],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,64],f16>
    %int1_84 = torch.constant.int 1
    %int2_85 = torch.constant.int 2
    %461 = torch.aten.transpose.int %460, %int1_84, %int2_85 : !torch.tensor<[1,12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int0_86 = torch.constant.int 0
    %462 = torch.aten.clone %461, %int0_86 : !torch.tensor<[1,77,12,64],f16>, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int1_87 = torch.constant.int 1
    %int77_88 = torch.constant.int 77
    %int768_89 = torch.constant.int 768
    %463 = torch.prim.ListConstruct %int1_87, %int77_88, %int768_89 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %464 = torch.aten._unsafe_view %462, %463 : !torch.tensor<[1,77,12,64],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %465 = torch.aten.t %385 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_90 = torch.constant.int 77
    %int768_91 = torch.constant.int 768
    %466 = torch.prim.ListConstruct %int77_90, %int768_91 : (!torch.int, !torch.int) -> !torch.list<int>
    %467 = torch.aten.view %464, %466 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_92 = torch.constant.int 1
    %int1_93 = torch.constant.int 1
    %468 = torch.aten.addmm %384, %467, %465, %int1_92, %int1_93 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_94 = torch.constant.int 1
    %int77_95 = torch.constant.int 77
    %int768_96 = torch.constant.int 768
    %469 = torch.prim.ListConstruct %int1_94, %int77_95, %int768_96 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %470 = torch.aten.view %468, %469 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_97 = torch.constant.int 1
    %471 = torch.aten.add.Tensor %403, %470, %int1_97 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_98 = torch.constant.int 768
    %472 = torch.prim.ListConstruct %int768_98 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_99 = torch.constant.float 1.000000e-05
    %result0_100, %result1_101, %result2_102 = torch.aten.native_layer_norm %471, %472, %383, %382, %float1.000000e-05_99 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %473 = torch.aten.t %381 : !torch.tensor -> !torch.tensor<[768,3072],f16>
    %int77_103 = torch.constant.int 77
    %int768_104 = torch.constant.int 768
    %474 = torch.prim.ListConstruct %int77_103, %int768_104 : (!torch.int, !torch.int) -> !torch.list<int>
    %475 = torch.aten.view %result0_100, %474 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_105 = torch.constant.int 1
    %int1_106 = torch.constant.int 1
    %476 = torch.aten.addmm %380, %475, %473, %int1_105, %int1_106 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,3072],f16>, !torch.int, !torch.int -> !torch.tensor<[77,3072],f16>
    %int1_107 = torch.constant.int 1
    %int77_108 = torch.constant.int 77
    %int3072 = torch.constant.int 3072
    %477 = torch.prim.ListConstruct %int1_107, %int77_108, %int3072 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %478 = torch.aten.view %476, %477 : !torch.tensor<[77,3072],f16>, !torch.list<int> -> !torch.tensor<[1,77,3072],f16>
    %479 = torch.tensor.literal(dense<1.702000e+00> : tensor<f64>) : !torch.tensor<[],f64>
    %480 = torch.aten.mul.Tensor %478, %479 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,3072],f16>
    %481 = torch.aten.sigmoid %480 : !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %482 = torch.aten.mul.Tensor %478, %481 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %483 = torch.aten.t %379 : !torch.tensor -> !torch.tensor<[3072,768],f16>
    %int77_109 = torch.constant.int 77
    %int3072_110 = torch.constant.int 3072
    %484 = torch.prim.ListConstruct %int77_109, %int3072_110 : (!torch.int, !torch.int) -> !torch.list<int>
    %485 = torch.aten.view %482, %484 : !torch.tensor<[1,77,3072],f16>, !torch.list<int> -> !torch.tensor<[77,3072],f16>
    %int1_111 = torch.constant.int 1
    %int1_112 = torch.constant.int 1
    %486 = torch.aten.addmm %378, %485, %483, %int1_111, %int1_112 : !torch.tensor, !torch.tensor<[77,3072],f16>, !torch.tensor<[3072,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_113 = torch.constant.int 1
    %int77_114 = torch.constant.int 77
    %int768_115 = torch.constant.int 768
    %487 = torch.prim.ListConstruct %int1_113, %int77_114, %int768_115 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %488 = torch.aten.view %486, %487 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_116 = torch.constant.int 1
    %489 = torch.aten.add.Tensor %471, %488, %int1_116 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_117 = torch.constant.int 768
    %490 = torch.prim.ListConstruct %int768_117 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_118 = torch.constant.float 1.000000e-05
    %result0_119, %result1_120, %result2_121 = torch.aten.native_layer_norm %489, %490, %377, %376, %float1.000000e-05_118 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %491 = torch.aten.t %375 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_122 = torch.constant.int 77
    %int768_123 = torch.constant.int 768
    %492 = torch.prim.ListConstruct %int77_122, %int768_123 : (!torch.int, !torch.int) -> !torch.list<int>
    %493 = torch.aten.view %result0_119, %492 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_124 = torch.constant.int 1
    %int1_125 = torch.constant.int 1
    %494 = torch.aten.addmm %374, %493, %491, %int1_124, %int1_125 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_126 = torch.constant.int 1
    %int77_127 = torch.constant.int 77
    %int768_128 = torch.constant.int 768
    %495 = torch.prim.ListConstruct %int1_126, %int77_127, %int768_128 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %496 = torch.aten.view %494, %495 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %497 = torch.tensor.literal(dense<1.250000e-01> : tensor<f64>) : !torch.tensor<[],f64>
    %498 = torch.aten.mul.Tensor %496, %497 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,768],f16>
    %499 = torch.aten.t %373 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_129 = torch.constant.int 77
    %int768_130 = torch.constant.int 768
    %500 = torch.prim.ListConstruct %int77_129, %int768_130 : (!torch.int, !torch.int) -> !torch.list<int>
    %501 = torch.aten.view %result0_119, %500 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_131 = torch.constant.int 1
    %int1_132 = torch.constant.int 1
    %502 = torch.aten.addmm %372, %501, %499, %int1_131, %int1_132 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_133 = torch.constant.int 1
    %int77_134 = torch.constant.int 77
    %int768_135 = torch.constant.int 768
    %503 = torch.prim.ListConstruct %int1_133, %int77_134, %int768_135 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %504 = torch.aten.view %502, %503 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_136 = torch.constant.int 1
    %int-1_137 = torch.constant.int -1
    %int12_138 = torch.constant.int 12
    %int64_139 = torch.constant.int 64
    %505 = torch.prim.ListConstruct %int1_136, %int-1_137, %int12_138, %int64_139 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %506 = torch.aten.view %504, %505 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_140 = torch.constant.int 1
    %int2_141 = torch.constant.int 2
    %507 = torch.aten.transpose.int %506, %int1_140, %int2_141 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_142 = torch.constant.int 0
    %508 = torch.aten.clone %507, %int0_142 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %509 = torch.aten.t %371 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_143 = torch.constant.int 77
    %int768_144 = torch.constant.int 768
    %510 = torch.prim.ListConstruct %int77_143, %int768_144 : (!torch.int, !torch.int) -> !torch.list<int>
    %511 = torch.aten.view %result0_119, %510 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_145 = torch.constant.int 1
    %int1_146 = torch.constant.int 1
    %512 = torch.aten.addmm %370, %511, %509, %int1_145, %int1_146 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_147 = torch.constant.int 1
    %int77_148 = torch.constant.int 77
    %int768_149 = torch.constant.int 768
    %513 = torch.prim.ListConstruct %int1_147, %int77_148, %int768_149 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %514 = torch.aten.view %512, %513 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_150 = torch.constant.int 1
    %int-1_151 = torch.constant.int -1
    %int12_152 = torch.constant.int 12
    %int64_153 = torch.constant.int 64
    %515 = torch.prim.ListConstruct %int1_150, %int-1_151, %int12_152, %int64_153 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %516 = torch.aten.view %514, %515 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_154 = torch.constant.int 1
    %int2_155 = torch.constant.int 2
    %517 = torch.aten.transpose.int %516, %int1_154, %int2_155 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_156 = torch.constant.int 0
    %518 = torch.aten.clone %517, %int0_156 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int1_157 = torch.constant.int 1
    %int77_158 = torch.constant.int 77
    %int12_159 = torch.constant.int 12
    %int64_160 = torch.constant.int 64
    %519 = torch.prim.ListConstruct %int1_157, %int77_158, %int12_159, %int64_160 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %520 = torch.aten.view %498, %519 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_161 = torch.constant.int 1
    %int2_162 = torch.constant.int 2
    %521 = torch.aten.transpose.int %520, %int1_161, %int2_162 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_163 = torch.constant.int 0
    %522 = torch.aten.clone %521, %int0_163 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int12_164 = torch.constant.int 12
    %int-1_165 = torch.constant.int -1
    %int64_166 = torch.constant.int 64
    %523 = torch.prim.ListConstruct %int12_164, %int-1_165, %int64_166 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %524 = torch.aten.view %522, %523 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_167 = torch.constant.int 12
    %int-1_168 = torch.constant.int -1
    %int64_169 = torch.constant.int 64
    %525 = torch.prim.ListConstruct %int12_167, %int-1_168, %int64_169 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %526 = torch.aten.view %508, %525 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_170 = torch.constant.int 12
    %int-1_171 = torch.constant.int -1
    %int64_172 = torch.constant.int 64
    %527 = torch.prim.ListConstruct %int12_170, %int-1_171, %int64_172 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %528 = torch.aten.view %518, %527 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int1_173 = torch.constant.int 1
    %int2_174 = torch.constant.int 2
    %529 = torch.aten.transpose.int %526, %int1_173, %int2_174 : !torch.tensor<[12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[12,64,77],f16>
    %530 = torch.aten.bmm %524, %529 : !torch.tensor<[12,77,64],f16>, !torch.tensor<[12,64,77],f16> -> !torch.tensor<[12,77,77],f16>
    %int1_175 = torch.constant.int 1
    %int12_176 = torch.constant.int 12
    %int77_177 = torch.constant.int 77
    %int77_178 = torch.constant.int 77
    %531 = torch.prim.ListConstruct %int1_175, %int12_176, %int77_177, %int77_178 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %532 = torch.aten.view %530, %531 : !torch.tensor<[12,77,77],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,77],f16>
    %int1_179 = torch.constant.int 1
    %533 = torch.aten.add.Tensor %532, %410, %int1_179 : !torch.tensor<[1,12,77,77],f16>, !torch.tensor<[1,1,77,77],f16>, !torch.int -> !torch.tensor<[1,12,77,77],f16>
    %int12_180 = torch.constant.int 12
    %int77_181 = torch.constant.int 77
    %int77_182 = torch.constant.int 77
    %534 = torch.prim.ListConstruct %int12_180, %int77_181, %int77_182 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %535 = torch.aten.view %533, %534 : !torch.tensor<[1,12,77,77],f16>, !torch.list<int> -> !torch.tensor<[12,77,77],f16>
    %int-1_183 = torch.constant.int -1
    %false_184 = torch.constant.bool false
    %536 = torch.aten._softmax %535, %int-1_183, %false_184 : !torch.tensor<[12,77,77],f16>, !torch.int, !torch.bool -> !torch.tensor<[12,77,77],f16>
    %537 = torch.aten.bmm %536, %528 : !torch.tensor<[12,77,77],f16>, !torch.tensor<[12,77,64],f16> -> !torch.tensor<[12,77,64],f16>
    %int1_185 = torch.constant.int 1
    %int12_186 = torch.constant.int 12
    %int77_187 = torch.constant.int 77
    %int64_188 = torch.constant.int 64
    %538 = torch.prim.ListConstruct %int1_185, %int12_186, %int77_187, %int64_188 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %539 = torch.aten.view %537, %538 : !torch.tensor<[12,77,64],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,64],f16>
    %int1_189 = torch.constant.int 1
    %int2_190 = torch.constant.int 2
    %540 = torch.aten.transpose.int %539, %int1_189, %int2_190 : !torch.tensor<[1,12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int0_191 = torch.constant.int 0
    %541 = torch.aten.clone %540, %int0_191 : !torch.tensor<[1,77,12,64],f16>, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int1_192 = torch.constant.int 1
    %int77_193 = torch.constant.int 77
    %int768_194 = torch.constant.int 768
    %542 = torch.prim.ListConstruct %int1_192, %int77_193, %int768_194 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %543 = torch.aten._unsafe_view %541, %542 : !torch.tensor<[1,77,12,64],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %544 = torch.aten.t %369 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_195 = torch.constant.int 77
    %int768_196 = torch.constant.int 768
    %545 = torch.prim.ListConstruct %int77_195, %int768_196 : (!torch.int, !torch.int) -> !torch.list<int>
    %546 = torch.aten.view %543, %545 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_197 = torch.constant.int 1
    %int1_198 = torch.constant.int 1
    %547 = torch.aten.addmm %368, %546, %544, %int1_197, %int1_198 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_199 = torch.constant.int 1
    %int77_200 = torch.constant.int 77
    %int768_201 = torch.constant.int 768
    %548 = torch.prim.ListConstruct %int1_199, %int77_200, %int768_201 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %549 = torch.aten.view %547, %548 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_202 = torch.constant.int 1
    %550 = torch.aten.add.Tensor %489, %549, %int1_202 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_203 = torch.constant.int 768
    %551 = torch.prim.ListConstruct %int768_203 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_204 = torch.constant.float 1.000000e-05
    %result0_205, %result1_206, %result2_207 = torch.aten.native_layer_norm %550, %551, %367, %366, %float1.000000e-05_204 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %552 = torch.aten.t %365 : !torch.tensor -> !torch.tensor<[768,3072],f16>
    %int77_208 = torch.constant.int 77
    %int768_209 = torch.constant.int 768
    %553 = torch.prim.ListConstruct %int77_208, %int768_209 : (!torch.int, !torch.int) -> !torch.list<int>
    %554 = torch.aten.view %result0_205, %553 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_210 = torch.constant.int 1
    %int1_211 = torch.constant.int 1
    %555 = torch.aten.addmm %364, %554, %552, %int1_210, %int1_211 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,3072],f16>, !torch.int, !torch.int -> !torch.tensor<[77,3072],f16>
    %int1_212 = torch.constant.int 1
    %int77_213 = torch.constant.int 77
    %int3072_214 = torch.constant.int 3072
    %556 = torch.prim.ListConstruct %int1_212, %int77_213, %int3072_214 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %557 = torch.aten.view %555, %556 : !torch.tensor<[77,3072],f16>, !torch.list<int> -> !torch.tensor<[1,77,3072],f16>
    %558 = torch.tensor.literal(dense<1.702000e+00> : tensor<f64>) : !torch.tensor<[],f64>
    %559 = torch.aten.mul.Tensor %557, %558 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,3072],f16>
    %560 = torch.aten.sigmoid %559 : !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %561 = torch.aten.mul.Tensor %557, %560 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %562 = torch.aten.t %363 : !torch.tensor -> !torch.tensor<[3072,768],f16>
    %int77_215 = torch.constant.int 77
    %int3072_216 = torch.constant.int 3072
    %563 = torch.prim.ListConstruct %int77_215, %int3072_216 : (!torch.int, !torch.int) -> !torch.list<int>
    %564 = torch.aten.view %561, %563 : !torch.tensor<[1,77,3072],f16>, !torch.list<int> -> !torch.tensor<[77,3072],f16>
    %int1_217 = torch.constant.int 1
    %int1_218 = torch.constant.int 1
    %565 = torch.aten.addmm %362, %564, %562, %int1_217, %int1_218 : !torch.tensor, !torch.tensor<[77,3072],f16>, !torch.tensor<[3072,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_219 = torch.constant.int 1
    %int77_220 = torch.constant.int 77
    %int768_221 = torch.constant.int 768
    %566 = torch.prim.ListConstruct %int1_219, %int77_220, %int768_221 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %567 = torch.aten.view %565, %566 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_222 = torch.constant.int 1
    %568 = torch.aten.add.Tensor %550, %567, %int1_222 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_223 = torch.constant.int 768
    %569 = torch.prim.ListConstruct %int768_223 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_224 = torch.constant.float 1.000000e-05
    %result0_225, %result1_226, %result2_227 = torch.aten.native_layer_norm %568, %569, %361, %360, %float1.000000e-05_224 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %570 = torch.aten.t %359 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_228 = torch.constant.int 77
    %int768_229 = torch.constant.int 768
    %571 = torch.prim.ListConstruct %int77_228, %int768_229 : (!torch.int, !torch.int) -> !torch.list<int>
    %572 = torch.aten.view %result0_225, %571 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_230 = torch.constant.int 1
    %int1_231 = torch.constant.int 1
    %573 = torch.aten.addmm %358, %572, %570, %int1_230, %int1_231 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_232 = torch.constant.int 1
    %int77_233 = torch.constant.int 77
    %int768_234 = torch.constant.int 768
    %574 = torch.prim.ListConstruct %int1_232, %int77_233, %int768_234 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %575 = torch.aten.view %573, %574 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %576 = torch.tensor.literal(dense<1.250000e-01> : tensor<f64>) : !torch.tensor<[],f64>
    %577 = torch.aten.mul.Tensor %575, %576 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,768],f16>
    %578 = torch.aten.t %357 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_235 = torch.constant.int 77
    %int768_236 = torch.constant.int 768
    %579 = torch.prim.ListConstruct %int77_235, %int768_236 : (!torch.int, !torch.int) -> !torch.list<int>
    %580 = torch.aten.view %result0_225, %579 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_237 = torch.constant.int 1
    %int1_238 = torch.constant.int 1
    %581 = torch.aten.addmm %356, %580, %578, %int1_237, %int1_238 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_239 = torch.constant.int 1
    %int77_240 = torch.constant.int 77
    %int768_241 = torch.constant.int 768
    %582 = torch.prim.ListConstruct %int1_239, %int77_240, %int768_241 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %583 = torch.aten.view %581, %582 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_242 = torch.constant.int 1
    %int-1_243 = torch.constant.int -1
    %int12_244 = torch.constant.int 12
    %int64_245 = torch.constant.int 64
    %584 = torch.prim.ListConstruct %int1_242, %int-1_243, %int12_244, %int64_245 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %585 = torch.aten.view %583, %584 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_246 = torch.constant.int 1
    %int2_247 = torch.constant.int 2
    %586 = torch.aten.transpose.int %585, %int1_246, %int2_247 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_248 = torch.constant.int 0
    %587 = torch.aten.clone %586, %int0_248 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %588 = torch.aten.t %355 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_249 = torch.constant.int 77
    %int768_250 = torch.constant.int 768
    %589 = torch.prim.ListConstruct %int77_249, %int768_250 : (!torch.int, !torch.int) -> !torch.list<int>
    %590 = torch.aten.view %result0_225, %589 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_251 = torch.constant.int 1
    %int1_252 = torch.constant.int 1
    %591 = torch.aten.addmm %354, %590, %588, %int1_251, %int1_252 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_253 = torch.constant.int 1
    %int77_254 = torch.constant.int 77
    %int768_255 = torch.constant.int 768
    %592 = torch.prim.ListConstruct %int1_253, %int77_254, %int768_255 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %593 = torch.aten.view %591, %592 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_256 = torch.constant.int 1
    %int-1_257 = torch.constant.int -1
    %int12_258 = torch.constant.int 12
    %int64_259 = torch.constant.int 64
    %594 = torch.prim.ListConstruct %int1_256, %int-1_257, %int12_258, %int64_259 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %595 = torch.aten.view %593, %594 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_260 = torch.constant.int 1
    %int2_261 = torch.constant.int 2
    %596 = torch.aten.transpose.int %595, %int1_260, %int2_261 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_262 = torch.constant.int 0
    %597 = torch.aten.clone %596, %int0_262 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int1_263 = torch.constant.int 1
    %int77_264 = torch.constant.int 77
    %int12_265 = torch.constant.int 12
    %int64_266 = torch.constant.int 64
    %598 = torch.prim.ListConstruct %int1_263, %int77_264, %int12_265, %int64_266 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %599 = torch.aten.view %577, %598 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_267 = torch.constant.int 1
    %int2_268 = torch.constant.int 2
    %600 = torch.aten.transpose.int %599, %int1_267, %int2_268 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_269 = torch.constant.int 0
    %601 = torch.aten.clone %600, %int0_269 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int12_270 = torch.constant.int 12
    %int-1_271 = torch.constant.int -1
    %int64_272 = torch.constant.int 64
    %602 = torch.prim.ListConstruct %int12_270, %int-1_271, %int64_272 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %603 = torch.aten.view %601, %602 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_273 = torch.constant.int 12
    %int-1_274 = torch.constant.int -1
    %int64_275 = torch.constant.int 64
    %604 = torch.prim.ListConstruct %int12_273, %int-1_274, %int64_275 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %605 = torch.aten.view %587, %604 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_276 = torch.constant.int 12
    %int-1_277 = torch.constant.int -1
    %int64_278 = torch.constant.int 64
    %606 = torch.prim.ListConstruct %int12_276, %int-1_277, %int64_278 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %607 = torch.aten.view %597, %606 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int1_279 = torch.constant.int 1
    %int2_280 = torch.constant.int 2
    %608 = torch.aten.transpose.int %605, %int1_279, %int2_280 : !torch.tensor<[12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[12,64,77],f16>
    %609 = torch.aten.bmm %603, %608 : !torch.tensor<[12,77,64],f16>, !torch.tensor<[12,64,77],f16> -> !torch.tensor<[12,77,77],f16>
    %int1_281 = torch.constant.int 1
    %int12_282 = torch.constant.int 12
    %int77_283 = torch.constant.int 77
    %int77_284 = torch.constant.int 77
    %610 = torch.prim.ListConstruct %int1_281, %int12_282, %int77_283, %int77_284 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %611 = torch.aten.view %609, %610 : !torch.tensor<[12,77,77],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,77],f16>
    %int1_285 = torch.constant.int 1
    %612 = torch.aten.add.Tensor %611, %410, %int1_285 : !torch.tensor<[1,12,77,77],f16>, !torch.tensor<[1,1,77,77],f16>, !torch.int -> !torch.tensor<[1,12,77,77],f16>
    %int12_286 = torch.constant.int 12
    %int77_287 = torch.constant.int 77
    %int77_288 = torch.constant.int 77
    %613 = torch.prim.ListConstruct %int12_286, %int77_287, %int77_288 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %614 = torch.aten.view %612, %613 : !torch.tensor<[1,12,77,77],f16>, !torch.list<int> -> !torch.tensor<[12,77,77],f16>
    %int-1_289 = torch.constant.int -1
    %false_290 = torch.constant.bool false
    %615 = torch.aten._softmax %614, %int-1_289, %false_290 : !torch.tensor<[12,77,77],f16>, !torch.int, !torch.bool -> !torch.tensor<[12,77,77],f16>
    %616 = torch.aten.bmm %615, %607 : !torch.tensor<[12,77,77],f16>, !torch.tensor<[12,77,64],f16> -> !torch.tensor<[12,77,64],f16>
    %int1_291 = torch.constant.int 1
    %int12_292 = torch.constant.int 12
    %int77_293 = torch.constant.int 77
    %int64_294 = torch.constant.int 64
    %617 = torch.prim.ListConstruct %int1_291, %int12_292, %int77_293, %int64_294 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %618 = torch.aten.view %616, %617 : !torch.tensor<[12,77,64],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,64],f16>
    %int1_295 = torch.constant.int 1
    %int2_296 = torch.constant.int 2
    %619 = torch.aten.transpose.int %618, %int1_295, %int2_296 : !torch.tensor<[1,12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int0_297 = torch.constant.int 0
    %620 = torch.aten.clone %619, %int0_297 : !torch.tensor<[1,77,12,64],f16>, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int1_298 = torch.constant.int 1
    %int77_299 = torch.constant.int 77
    %int768_300 = torch.constant.int 768
    %621 = torch.prim.ListConstruct %int1_298, %int77_299, %int768_300 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %622 = torch.aten._unsafe_view %620, %621 : !torch.tensor<[1,77,12,64],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %623 = torch.aten.t %353 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_301 = torch.constant.int 77
    %int768_302 = torch.constant.int 768
    %624 = torch.prim.ListConstruct %int77_301, %int768_302 : (!torch.int, !torch.int) -> !torch.list<int>
    %625 = torch.aten.view %622, %624 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_303 = torch.constant.int 1
    %int1_304 = torch.constant.int 1
    %626 = torch.aten.addmm %352, %625, %623, %int1_303, %int1_304 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_305 = torch.constant.int 1
    %int77_306 = torch.constant.int 77
    %int768_307 = torch.constant.int 768
    %627 = torch.prim.ListConstruct %int1_305, %int77_306, %int768_307 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %628 = torch.aten.view %626, %627 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_308 = torch.constant.int 1
    %629 = torch.aten.add.Tensor %568, %628, %int1_308 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_309 = torch.constant.int 768
    %630 = torch.prim.ListConstruct %int768_309 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_310 = torch.constant.float 1.000000e-05
    %result0_311, %result1_312, %result2_313 = torch.aten.native_layer_norm %629, %630, %351, %350, %float1.000000e-05_310 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %631 = torch.aten.t %349 : !torch.tensor -> !torch.tensor<[768,3072],f16>
    %int77_314 = torch.constant.int 77
    %int768_315 = torch.constant.int 768
    %632 = torch.prim.ListConstruct %int77_314, %int768_315 : (!torch.int, !torch.int) -> !torch.list<int>
    %633 = torch.aten.view %result0_311, %632 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_316 = torch.constant.int 1
    %int1_317 = torch.constant.int 1
    %634 = torch.aten.addmm %348, %633, %631, %int1_316, %int1_317 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,3072],f16>, !torch.int, !torch.int -> !torch.tensor<[77,3072],f16>
    %int1_318 = torch.constant.int 1
    %int77_319 = torch.constant.int 77
    %int3072_320 = torch.constant.int 3072
    %635 = torch.prim.ListConstruct %int1_318, %int77_319, %int3072_320 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %636 = torch.aten.view %634, %635 : !torch.tensor<[77,3072],f16>, !torch.list<int> -> !torch.tensor<[1,77,3072],f16>
    %637 = torch.tensor.literal(dense<1.702000e+00> : tensor<f64>) : !torch.tensor<[],f64>
    %638 = torch.aten.mul.Tensor %636, %637 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,3072],f16>
    %639 = torch.aten.sigmoid %638 : !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %640 = torch.aten.mul.Tensor %636, %639 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %641 = torch.aten.t %347 : !torch.tensor -> !torch.tensor<[3072,768],f16>
    %int77_321 = torch.constant.int 77
    %int3072_322 = torch.constant.int 3072
    %642 = torch.prim.ListConstruct %int77_321, %int3072_322 : (!torch.int, !torch.int) -> !torch.list<int>
    %643 = torch.aten.view %640, %642 : !torch.tensor<[1,77,3072],f16>, !torch.list<int> -> !torch.tensor<[77,3072],f16>
    %int1_323 = torch.constant.int 1
    %int1_324 = torch.constant.int 1
    %644 = torch.aten.addmm %346, %643, %641, %int1_323, %int1_324 : !torch.tensor, !torch.tensor<[77,3072],f16>, !torch.tensor<[3072,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_325 = torch.constant.int 1
    %int77_326 = torch.constant.int 77
    %int768_327 = torch.constant.int 768
    %645 = torch.prim.ListConstruct %int1_325, %int77_326, %int768_327 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %646 = torch.aten.view %644, %645 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_328 = torch.constant.int 1
    %647 = torch.aten.add.Tensor %629, %646, %int1_328 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_329 = torch.constant.int 768
    %648 = torch.prim.ListConstruct %int768_329 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_330 = torch.constant.float 1.000000e-05
    %result0_331, %result1_332, %result2_333 = torch.aten.native_layer_norm %647, %648, %345, %344, %float1.000000e-05_330 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %649 = torch.aten.t %343 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_334 = torch.constant.int 77
    %int768_335 = torch.constant.int 768
    %650 = torch.prim.ListConstruct %int77_334, %int768_335 : (!torch.int, !torch.int) -> !torch.list<int>
    %651 = torch.aten.view %result0_331, %650 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_336 = torch.constant.int 1
    %int1_337 = torch.constant.int 1
    %652 = torch.aten.addmm %342, %651, %649, %int1_336, %int1_337 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_338 = torch.constant.int 1
    %int77_339 = torch.constant.int 77
    %int768_340 = torch.constant.int 768
    %653 = torch.prim.ListConstruct %int1_338, %int77_339, %int768_340 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %654 = torch.aten.view %652, %653 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %655 = torch.tensor.literal(dense<1.250000e-01> : tensor<f64>) : !torch.tensor<[],f64>
    %656 = torch.aten.mul.Tensor %654, %655 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,768],f16>
    %657 = torch.aten.t %341 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_341 = torch.constant.int 77
    %int768_342 = torch.constant.int 768
    %658 = torch.prim.ListConstruct %int77_341, %int768_342 : (!torch.int, !torch.int) -> !torch.list<int>
    %659 = torch.aten.view %result0_331, %658 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_343 = torch.constant.int 1
    %int1_344 = torch.constant.int 1
    %660 = torch.aten.addmm %340, %659, %657, %int1_343, %int1_344 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_345 = torch.constant.int 1
    %int77_346 = torch.constant.int 77
    %int768_347 = torch.constant.int 768
    %661 = torch.prim.ListConstruct %int1_345, %int77_346, %int768_347 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %662 = torch.aten.view %660, %661 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_348 = torch.constant.int 1
    %int-1_349 = torch.constant.int -1
    %int12_350 = torch.constant.int 12
    %int64_351 = torch.constant.int 64
    %663 = torch.prim.ListConstruct %int1_348, %int-1_349, %int12_350, %int64_351 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %664 = torch.aten.view %662, %663 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_352 = torch.constant.int 1
    %int2_353 = torch.constant.int 2
    %665 = torch.aten.transpose.int %664, %int1_352, %int2_353 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_354 = torch.constant.int 0
    %666 = torch.aten.clone %665, %int0_354 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %667 = torch.aten.t %339 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_355 = torch.constant.int 77
    %int768_356 = torch.constant.int 768
    %668 = torch.prim.ListConstruct %int77_355, %int768_356 : (!torch.int, !torch.int) -> !torch.list<int>
    %669 = torch.aten.view %result0_331, %668 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_357 = torch.constant.int 1
    %int1_358 = torch.constant.int 1
    %670 = torch.aten.addmm %338, %669, %667, %int1_357, %int1_358 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_359 = torch.constant.int 1
    %int77_360 = torch.constant.int 77
    %int768_361 = torch.constant.int 768
    %671 = torch.prim.ListConstruct %int1_359, %int77_360, %int768_361 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %672 = torch.aten.view %670, %671 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_362 = torch.constant.int 1
    %int-1_363 = torch.constant.int -1
    %int12_364 = torch.constant.int 12
    %int64_365 = torch.constant.int 64
    %673 = torch.prim.ListConstruct %int1_362, %int-1_363, %int12_364, %int64_365 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %674 = torch.aten.view %672, %673 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_366 = torch.constant.int 1
    %int2_367 = torch.constant.int 2
    %675 = torch.aten.transpose.int %674, %int1_366, %int2_367 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_368 = torch.constant.int 0
    %676 = torch.aten.clone %675, %int0_368 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int1_369 = torch.constant.int 1
    %int77_370 = torch.constant.int 77
    %int12_371 = torch.constant.int 12
    %int64_372 = torch.constant.int 64
    %677 = torch.prim.ListConstruct %int1_369, %int77_370, %int12_371, %int64_372 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %678 = torch.aten.view %656, %677 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_373 = torch.constant.int 1
    %int2_374 = torch.constant.int 2
    %679 = torch.aten.transpose.int %678, %int1_373, %int2_374 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_375 = torch.constant.int 0
    %680 = torch.aten.clone %679, %int0_375 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int12_376 = torch.constant.int 12
    %int-1_377 = torch.constant.int -1
    %int64_378 = torch.constant.int 64
    %681 = torch.prim.ListConstruct %int12_376, %int-1_377, %int64_378 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %682 = torch.aten.view %680, %681 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_379 = torch.constant.int 12
    %int-1_380 = torch.constant.int -1
    %int64_381 = torch.constant.int 64
    %683 = torch.prim.ListConstruct %int12_379, %int-1_380, %int64_381 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %684 = torch.aten.view %666, %683 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_382 = torch.constant.int 12
    %int-1_383 = torch.constant.int -1
    %int64_384 = torch.constant.int 64
    %685 = torch.prim.ListConstruct %int12_382, %int-1_383, %int64_384 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %686 = torch.aten.view %676, %685 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int1_385 = torch.constant.int 1
    %int2_386 = torch.constant.int 2
    %687 = torch.aten.transpose.int %684, %int1_385, %int2_386 : !torch.tensor<[12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[12,64,77],f16>
    %688 = torch.aten.bmm %682, %687 : !torch.tensor<[12,77,64],f16>, !torch.tensor<[12,64,77],f16> -> !torch.tensor<[12,77,77],f16>
    %int1_387 = torch.constant.int 1
    %int12_388 = torch.constant.int 12
    %int77_389 = torch.constant.int 77
    %int77_390 = torch.constant.int 77
    %689 = torch.prim.ListConstruct %int1_387, %int12_388, %int77_389, %int77_390 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %690 = torch.aten.view %688, %689 : !torch.tensor<[12,77,77],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,77],f16>
    %int1_391 = torch.constant.int 1
    %691 = torch.aten.add.Tensor %690, %410, %int1_391 : !torch.tensor<[1,12,77,77],f16>, !torch.tensor<[1,1,77,77],f16>, !torch.int -> !torch.tensor<[1,12,77,77],f16>
    %int12_392 = torch.constant.int 12
    %int77_393 = torch.constant.int 77
    %int77_394 = torch.constant.int 77
    %692 = torch.prim.ListConstruct %int12_392, %int77_393, %int77_394 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %693 = torch.aten.view %691, %692 : !torch.tensor<[1,12,77,77],f16>, !torch.list<int> -> !torch.tensor<[12,77,77],f16>
    %int-1_395 = torch.constant.int -1
    %false_396 = torch.constant.bool false
    %694 = torch.aten._softmax %693, %int-1_395, %false_396 : !torch.tensor<[12,77,77],f16>, !torch.int, !torch.bool -> !torch.tensor<[12,77,77],f16>
    %695 = torch.aten.bmm %694, %686 : !torch.tensor<[12,77,77],f16>, !torch.tensor<[12,77,64],f16> -> !torch.tensor<[12,77,64],f16>
    %int1_397 = torch.constant.int 1
    %int12_398 = torch.constant.int 12
    %int77_399 = torch.constant.int 77
    %int64_400 = torch.constant.int 64
    %696 = torch.prim.ListConstruct %int1_397, %int12_398, %int77_399, %int64_400 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %697 = torch.aten.view %695, %696 : !torch.tensor<[12,77,64],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,64],f16>
    %int1_401 = torch.constant.int 1
    %int2_402 = torch.constant.int 2
    %698 = torch.aten.transpose.int %697, %int1_401, %int2_402 : !torch.tensor<[1,12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int0_403 = torch.constant.int 0
    %699 = torch.aten.clone %698, %int0_403 : !torch.tensor<[1,77,12,64],f16>, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int1_404 = torch.constant.int 1
    %int77_405 = torch.constant.int 77
    %int768_406 = torch.constant.int 768
    %700 = torch.prim.ListConstruct %int1_404, %int77_405, %int768_406 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %701 = torch.aten._unsafe_view %699, %700 : !torch.tensor<[1,77,12,64],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %702 = torch.aten.t %337 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_407 = torch.constant.int 77
    %int768_408 = torch.constant.int 768
    %703 = torch.prim.ListConstruct %int77_407, %int768_408 : (!torch.int, !torch.int) -> !torch.list<int>
    %704 = torch.aten.view %701, %703 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_409 = torch.constant.int 1
    %int1_410 = torch.constant.int 1
    %705 = torch.aten.addmm %336, %704, %702, %int1_409, %int1_410 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_411 = torch.constant.int 1
    %int77_412 = torch.constant.int 77
    %int768_413 = torch.constant.int 768
    %706 = torch.prim.ListConstruct %int1_411, %int77_412, %int768_413 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %707 = torch.aten.view %705, %706 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_414 = torch.constant.int 1
    %708 = torch.aten.add.Tensor %647, %707, %int1_414 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_415 = torch.constant.int 768
    %709 = torch.prim.ListConstruct %int768_415 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_416 = torch.constant.float 1.000000e-05
    %result0_417, %result1_418, %result2_419 = torch.aten.native_layer_norm %708, %709, %335, %334, %float1.000000e-05_416 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %710 = torch.aten.t %333 : !torch.tensor -> !torch.tensor<[768,3072],f16>
    %int77_420 = torch.constant.int 77
    %int768_421 = torch.constant.int 768
    %711 = torch.prim.ListConstruct %int77_420, %int768_421 : (!torch.int, !torch.int) -> !torch.list<int>
    %712 = torch.aten.view %result0_417, %711 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_422 = torch.constant.int 1
    %int1_423 = torch.constant.int 1
    %713 = torch.aten.addmm %332, %712, %710, %int1_422, %int1_423 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,3072],f16>, !torch.int, !torch.int -> !torch.tensor<[77,3072],f16>
    %int1_424 = torch.constant.int 1
    %int77_425 = torch.constant.int 77
    %int3072_426 = torch.constant.int 3072
    %714 = torch.prim.ListConstruct %int1_424, %int77_425, %int3072_426 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %715 = torch.aten.view %713, %714 : !torch.tensor<[77,3072],f16>, !torch.list<int> -> !torch.tensor<[1,77,3072],f16>
    %716 = torch.tensor.literal(dense<1.702000e+00> : tensor<f64>) : !torch.tensor<[],f64>
    %717 = torch.aten.mul.Tensor %715, %716 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,3072],f16>
    %718 = torch.aten.sigmoid %717 : !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %719 = torch.aten.mul.Tensor %715, %718 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %720 = torch.aten.t %331 : !torch.tensor -> !torch.tensor<[3072,768],f16>
    %int77_427 = torch.constant.int 77
    %int3072_428 = torch.constant.int 3072
    %721 = torch.prim.ListConstruct %int77_427, %int3072_428 : (!torch.int, !torch.int) -> !torch.list<int>
    %722 = torch.aten.view %719, %721 : !torch.tensor<[1,77,3072],f16>, !torch.list<int> -> !torch.tensor<[77,3072],f16>
    %int1_429 = torch.constant.int 1
    %int1_430 = torch.constant.int 1
    %723 = torch.aten.addmm %330, %722, %720, %int1_429, %int1_430 : !torch.tensor, !torch.tensor<[77,3072],f16>, !torch.tensor<[3072,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_431 = torch.constant.int 1
    %int77_432 = torch.constant.int 77
    %int768_433 = torch.constant.int 768
    %724 = torch.prim.ListConstruct %int1_431, %int77_432, %int768_433 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %725 = torch.aten.view %723, %724 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_434 = torch.constant.int 1
    %726 = torch.aten.add.Tensor %708, %725, %int1_434 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_435 = torch.constant.int 768
    %727 = torch.prim.ListConstruct %int768_435 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_436 = torch.constant.float 1.000000e-05
    %result0_437, %result1_438, %result2_439 = torch.aten.native_layer_norm %726, %727, %329, %328, %float1.000000e-05_436 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %728 = torch.aten.t %327 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_440 = torch.constant.int 77
    %int768_441 = torch.constant.int 768
    %729 = torch.prim.ListConstruct %int77_440, %int768_441 : (!torch.int, !torch.int) -> !torch.list<int>
    %730 = torch.aten.view %result0_437, %729 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_442 = torch.constant.int 1
    %int1_443 = torch.constant.int 1
    %731 = torch.aten.addmm %326, %730, %728, %int1_442, %int1_443 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_444 = torch.constant.int 1
    %int77_445 = torch.constant.int 77
    %int768_446 = torch.constant.int 768
    %732 = torch.prim.ListConstruct %int1_444, %int77_445, %int768_446 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %733 = torch.aten.view %731, %732 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %734 = torch.tensor.literal(dense<1.250000e-01> : tensor<f64>) : !torch.tensor<[],f64>
    %735 = torch.aten.mul.Tensor %733, %734 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,768],f16>
    %736 = torch.aten.t %325 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_447 = torch.constant.int 77
    %int768_448 = torch.constant.int 768
    %737 = torch.prim.ListConstruct %int77_447, %int768_448 : (!torch.int, !torch.int) -> !torch.list<int>
    %738 = torch.aten.view %result0_437, %737 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_449 = torch.constant.int 1
    %int1_450 = torch.constant.int 1
    %739 = torch.aten.addmm %324, %738, %736, %int1_449, %int1_450 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_451 = torch.constant.int 1
    %int77_452 = torch.constant.int 77
    %int768_453 = torch.constant.int 768
    %740 = torch.prim.ListConstruct %int1_451, %int77_452, %int768_453 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %741 = torch.aten.view %739, %740 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_454 = torch.constant.int 1
    %int-1_455 = torch.constant.int -1
    %int12_456 = torch.constant.int 12
    %int64_457 = torch.constant.int 64
    %742 = torch.prim.ListConstruct %int1_454, %int-1_455, %int12_456, %int64_457 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %743 = torch.aten.view %741, %742 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_458 = torch.constant.int 1
    %int2_459 = torch.constant.int 2
    %744 = torch.aten.transpose.int %743, %int1_458, %int2_459 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_460 = torch.constant.int 0
    %745 = torch.aten.clone %744, %int0_460 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %746 = torch.aten.t %323 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_461 = torch.constant.int 77
    %int768_462 = torch.constant.int 768
    %747 = torch.prim.ListConstruct %int77_461, %int768_462 : (!torch.int, !torch.int) -> !torch.list<int>
    %748 = torch.aten.view %result0_437, %747 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_463 = torch.constant.int 1
    %int1_464 = torch.constant.int 1
    %749 = torch.aten.addmm %322, %748, %746, %int1_463, %int1_464 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_465 = torch.constant.int 1
    %int77_466 = torch.constant.int 77
    %int768_467 = torch.constant.int 768
    %750 = torch.prim.ListConstruct %int1_465, %int77_466, %int768_467 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %751 = torch.aten.view %749, %750 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_468 = torch.constant.int 1
    %int-1_469 = torch.constant.int -1
    %int12_470 = torch.constant.int 12
    %int64_471 = torch.constant.int 64
    %752 = torch.prim.ListConstruct %int1_468, %int-1_469, %int12_470, %int64_471 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %753 = torch.aten.view %751, %752 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_472 = torch.constant.int 1
    %int2_473 = torch.constant.int 2
    %754 = torch.aten.transpose.int %753, %int1_472, %int2_473 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_474 = torch.constant.int 0
    %755 = torch.aten.clone %754, %int0_474 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int1_475 = torch.constant.int 1
    %int77_476 = torch.constant.int 77
    %int12_477 = torch.constant.int 12
    %int64_478 = torch.constant.int 64
    %756 = torch.prim.ListConstruct %int1_475, %int77_476, %int12_477, %int64_478 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %757 = torch.aten.view %735, %756 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_479 = torch.constant.int 1
    %int2_480 = torch.constant.int 2
    %758 = torch.aten.transpose.int %757, %int1_479, %int2_480 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_481 = torch.constant.int 0
    %759 = torch.aten.clone %758, %int0_481 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int12_482 = torch.constant.int 12
    %int-1_483 = torch.constant.int -1
    %int64_484 = torch.constant.int 64
    %760 = torch.prim.ListConstruct %int12_482, %int-1_483, %int64_484 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %761 = torch.aten.view %759, %760 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_485 = torch.constant.int 12
    %int-1_486 = torch.constant.int -1
    %int64_487 = torch.constant.int 64
    %762 = torch.prim.ListConstruct %int12_485, %int-1_486, %int64_487 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %763 = torch.aten.view %745, %762 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_488 = torch.constant.int 12
    %int-1_489 = torch.constant.int -1
    %int64_490 = torch.constant.int 64
    %764 = torch.prim.ListConstruct %int12_488, %int-1_489, %int64_490 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %765 = torch.aten.view %755, %764 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int1_491 = torch.constant.int 1
    %int2_492 = torch.constant.int 2
    %766 = torch.aten.transpose.int %763, %int1_491, %int2_492 : !torch.tensor<[12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[12,64,77],f16>
    %767 = torch.aten.bmm %761, %766 : !torch.tensor<[12,77,64],f16>, !torch.tensor<[12,64,77],f16> -> !torch.tensor<[12,77,77],f16>
    %int1_493 = torch.constant.int 1
    %int12_494 = torch.constant.int 12
    %int77_495 = torch.constant.int 77
    %int77_496 = torch.constant.int 77
    %768 = torch.prim.ListConstruct %int1_493, %int12_494, %int77_495, %int77_496 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %769 = torch.aten.view %767, %768 : !torch.tensor<[12,77,77],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,77],f16>
    %int1_497 = torch.constant.int 1
    %770 = torch.aten.add.Tensor %769, %410, %int1_497 : !torch.tensor<[1,12,77,77],f16>, !torch.tensor<[1,1,77,77],f16>, !torch.int -> !torch.tensor<[1,12,77,77],f16>
    %int12_498 = torch.constant.int 12
    %int77_499 = torch.constant.int 77
    %int77_500 = torch.constant.int 77
    %771 = torch.prim.ListConstruct %int12_498, %int77_499, %int77_500 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %772 = torch.aten.view %770, %771 : !torch.tensor<[1,12,77,77],f16>, !torch.list<int> -> !torch.tensor<[12,77,77],f16>
    %int-1_501 = torch.constant.int -1
    %false_502 = torch.constant.bool false
    %773 = torch.aten._softmax %772, %int-1_501, %false_502 : !torch.tensor<[12,77,77],f16>, !torch.int, !torch.bool -> !torch.tensor<[12,77,77],f16>
    %774 = torch.aten.bmm %773, %765 : !torch.tensor<[12,77,77],f16>, !torch.tensor<[12,77,64],f16> -> !torch.tensor<[12,77,64],f16>
    %int1_503 = torch.constant.int 1
    %int12_504 = torch.constant.int 12
    %int77_505 = torch.constant.int 77
    %int64_506 = torch.constant.int 64
    %775 = torch.prim.ListConstruct %int1_503, %int12_504, %int77_505, %int64_506 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %776 = torch.aten.view %774, %775 : !torch.tensor<[12,77,64],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,64],f16>
    %int1_507 = torch.constant.int 1
    %int2_508 = torch.constant.int 2
    %777 = torch.aten.transpose.int %776, %int1_507, %int2_508 : !torch.tensor<[1,12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int0_509 = torch.constant.int 0
    %778 = torch.aten.clone %777, %int0_509 : !torch.tensor<[1,77,12,64],f16>, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int1_510 = torch.constant.int 1
    %int77_511 = torch.constant.int 77
    %int768_512 = torch.constant.int 768
    %779 = torch.prim.ListConstruct %int1_510, %int77_511, %int768_512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %780 = torch.aten._unsafe_view %778, %779 : !torch.tensor<[1,77,12,64],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %781 = torch.aten.t %321 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_513 = torch.constant.int 77
    %int768_514 = torch.constant.int 768
    %782 = torch.prim.ListConstruct %int77_513, %int768_514 : (!torch.int, !torch.int) -> !torch.list<int>
    %783 = torch.aten.view %780, %782 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_515 = torch.constant.int 1
    %int1_516 = torch.constant.int 1
    %784 = torch.aten.addmm %320, %783, %781, %int1_515, %int1_516 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_517 = torch.constant.int 1
    %int77_518 = torch.constant.int 77
    %int768_519 = torch.constant.int 768
    %785 = torch.prim.ListConstruct %int1_517, %int77_518, %int768_519 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %786 = torch.aten.view %784, %785 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_520 = torch.constant.int 1
    %787 = torch.aten.add.Tensor %726, %786, %int1_520 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_521 = torch.constant.int 768
    %788 = torch.prim.ListConstruct %int768_521 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_522 = torch.constant.float 1.000000e-05
    %result0_523, %result1_524, %result2_525 = torch.aten.native_layer_norm %787, %788, %319, %318, %float1.000000e-05_522 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %789 = torch.aten.t %317 : !torch.tensor -> !torch.tensor<[768,3072],f16>
    %int77_526 = torch.constant.int 77
    %int768_527 = torch.constant.int 768
    %790 = torch.prim.ListConstruct %int77_526, %int768_527 : (!torch.int, !torch.int) -> !torch.list<int>
    %791 = torch.aten.view %result0_523, %790 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_528 = torch.constant.int 1
    %int1_529 = torch.constant.int 1
    %792 = torch.aten.addmm %316, %791, %789, %int1_528, %int1_529 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,3072],f16>, !torch.int, !torch.int -> !torch.tensor<[77,3072],f16>
    %int1_530 = torch.constant.int 1
    %int77_531 = torch.constant.int 77
    %int3072_532 = torch.constant.int 3072
    %793 = torch.prim.ListConstruct %int1_530, %int77_531, %int3072_532 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %794 = torch.aten.view %792, %793 : !torch.tensor<[77,3072],f16>, !torch.list<int> -> !torch.tensor<[1,77,3072],f16>
    %795 = torch.tensor.literal(dense<1.702000e+00> : tensor<f64>) : !torch.tensor<[],f64>
    %796 = torch.aten.mul.Tensor %794, %795 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,3072],f16>
    %797 = torch.aten.sigmoid %796 : !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %798 = torch.aten.mul.Tensor %794, %797 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %799 = torch.aten.t %315 : !torch.tensor -> !torch.tensor<[3072,768],f16>
    %int77_533 = torch.constant.int 77
    %int3072_534 = torch.constant.int 3072
    %800 = torch.prim.ListConstruct %int77_533, %int3072_534 : (!torch.int, !torch.int) -> !torch.list<int>
    %801 = torch.aten.view %798, %800 : !torch.tensor<[1,77,3072],f16>, !torch.list<int> -> !torch.tensor<[77,3072],f16>
    %int1_535 = torch.constant.int 1
    %int1_536 = torch.constant.int 1
    %802 = torch.aten.addmm %314, %801, %799, %int1_535, %int1_536 : !torch.tensor, !torch.tensor<[77,3072],f16>, !torch.tensor<[3072,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_537 = torch.constant.int 1
    %int77_538 = torch.constant.int 77
    %int768_539 = torch.constant.int 768
    %803 = torch.prim.ListConstruct %int1_537, %int77_538, %int768_539 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %804 = torch.aten.view %802, %803 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_540 = torch.constant.int 1
    %805 = torch.aten.add.Tensor %787, %804, %int1_540 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_541 = torch.constant.int 768
    %806 = torch.prim.ListConstruct %int768_541 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_542 = torch.constant.float 1.000000e-05
    %result0_543, %result1_544, %result2_545 = torch.aten.native_layer_norm %805, %806, %313, %312, %float1.000000e-05_542 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %807 = torch.aten.t %311 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_546 = torch.constant.int 77
    %int768_547 = torch.constant.int 768
    %808 = torch.prim.ListConstruct %int77_546, %int768_547 : (!torch.int, !torch.int) -> !torch.list<int>
    %809 = torch.aten.view %result0_543, %808 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_548 = torch.constant.int 1
    %int1_549 = torch.constant.int 1
    %810 = torch.aten.addmm %310, %809, %807, %int1_548, %int1_549 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_550 = torch.constant.int 1
    %int77_551 = torch.constant.int 77
    %int768_552 = torch.constant.int 768
    %811 = torch.prim.ListConstruct %int1_550, %int77_551, %int768_552 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %812 = torch.aten.view %810, %811 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %813 = torch.tensor.literal(dense<1.250000e-01> : tensor<f64>) : !torch.tensor<[],f64>
    %814 = torch.aten.mul.Tensor %812, %813 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,768],f16>
    %815 = torch.aten.t %309 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_553 = torch.constant.int 77
    %int768_554 = torch.constant.int 768
    %816 = torch.prim.ListConstruct %int77_553, %int768_554 : (!torch.int, !torch.int) -> !torch.list<int>
    %817 = torch.aten.view %result0_543, %816 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_555 = torch.constant.int 1
    %int1_556 = torch.constant.int 1
    %818 = torch.aten.addmm %308, %817, %815, %int1_555, %int1_556 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_557 = torch.constant.int 1
    %int77_558 = torch.constant.int 77
    %int768_559 = torch.constant.int 768
    %819 = torch.prim.ListConstruct %int1_557, %int77_558, %int768_559 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %820 = torch.aten.view %818, %819 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_560 = torch.constant.int 1
    %int-1_561 = torch.constant.int -1
    %int12_562 = torch.constant.int 12
    %int64_563 = torch.constant.int 64
    %821 = torch.prim.ListConstruct %int1_560, %int-1_561, %int12_562, %int64_563 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %822 = torch.aten.view %820, %821 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_564 = torch.constant.int 1
    %int2_565 = torch.constant.int 2
    %823 = torch.aten.transpose.int %822, %int1_564, %int2_565 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_566 = torch.constant.int 0
    %824 = torch.aten.clone %823, %int0_566 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %825 = torch.aten.t %307 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_567 = torch.constant.int 77
    %int768_568 = torch.constant.int 768
    %826 = torch.prim.ListConstruct %int77_567, %int768_568 : (!torch.int, !torch.int) -> !torch.list<int>
    %827 = torch.aten.view %result0_543, %826 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_569 = torch.constant.int 1
    %int1_570 = torch.constant.int 1
    %828 = torch.aten.addmm %306, %827, %825, %int1_569, %int1_570 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_571 = torch.constant.int 1
    %int77_572 = torch.constant.int 77
    %int768_573 = torch.constant.int 768
    %829 = torch.prim.ListConstruct %int1_571, %int77_572, %int768_573 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %830 = torch.aten.view %828, %829 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_574 = torch.constant.int 1
    %int-1_575 = torch.constant.int -1
    %int12_576 = torch.constant.int 12
    %int64_577 = torch.constant.int 64
    %831 = torch.prim.ListConstruct %int1_574, %int-1_575, %int12_576, %int64_577 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %832 = torch.aten.view %830, %831 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_578 = torch.constant.int 1
    %int2_579 = torch.constant.int 2
    %833 = torch.aten.transpose.int %832, %int1_578, %int2_579 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_580 = torch.constant.int 0
    %834 = torch.aten.clone %833, %int0_580 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int1_581 = torch.constant.int 1
    %int77_582 = torch.constant.int 77
    %int12_583 = torch.constant.int 12
    %int64_584 = torch.constant.int 64
    %835 = torch.prim.ListConstruct %int1_581, %int77_582, %int12_583, %int64_584 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %836 = torch.aten.view %814, %835 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_585 = torch.constant.int 1
    %int2_586 = torch.constant.int 2
    %837 = torch.aten.transpose.int %836, %int1_585, %int2_586 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_587 = torch.constant.int 0
    %838 = torch.aten.clone %837, %int0_587 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int12_588 = torch.constant.int 12
    %int-1_589 = torch.constant.int -1
    %int64_590 = torch.constant.int 64
    %839 = torch.prim.ListConstruct %int12_588, %int-1_589, %int64_590 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %840 = torch.aten.view %838, %839 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_591 = torch.constant.int 12
    %int-1_592 = torch.constant.int -1
    %int64_593 = torch.constant.int 64
    %841 = torch.prim.ListConstruct %int12_591, %int-1_592, %int64_593 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %842 = torch.aten.view %824, %841 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_594 = torch.constant.int 12
    %int-1_595 = torch.constant.int -1
    %int64_596 = torch.constant.int 64
    %843 = torch.prim.ListConstruct %int12_594, %int-1_595, %int64_596 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %844 = torch.aten.view %834, %843 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int1_597 = torch.constant.int 1
    %int2_598 = torch.constant.int 2
    %845 = torch.aten.transpose.int %842, %int1_597, %int2_598 : !torch.tensor<[12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[12,64,77],f16>
    %846 = torch.aten.bmm %840, %845 : !torch.tensor<[12,77,64],f16>, !torch.tensor<[12,64,77],f16> -> !torch.tensor<[12,77,77],f16>
    %int1_599 = torch.constant.int 1
    %int12_600 = torch.constant.int 12
    %int77_601 = torch.constant.int 77
    %int77_602 = torch.constant.int 77
    %847 = torch.prim.ListConstruct %int1_599, %int12_600, %int77_601, %int77_602 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %848 = torch.aten.view %846, %847 : !torch.tensor<[12,77,77],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,77],f16>
    %int1_603 = torch.constant.int 1
    %849 = torch.aten.add.Tensor %848, %410, %int1_603 : !torch.tensor<[1,12,77,77],f16>, !torch.tensor<[1,1,77,77],f16>, !torch.int -> !torch.tensor<[1,12,77,77],f16>
    %int12_604 = torch.constant.int 12
    %int77_605 = torch.constant.int 77
    %int77_606 = torch.constant.int 77
    %850 = torch.prim.ListConstruct %int12_604, %int77_605, %int77_606 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %851 = torch.aten.view %849, %850 : !torch.tensor<[1,12,77,77],f16>, !torch.list<int> -> !torch.tensor<[12,77,77],f16>
    %int-1_607 = torch.constant.int -1
    %false_608 = torch.constant.bool false
    %852 = torch.aten._softmax %851, %int-1_607, %false_608 : !torch.tensor<[12,77,77],f16>, !torch.int, !torch.bool -> !torch.tensor<[12,77,77],f16>
    %853 = torch.aten.bmm %852, %844 : !torch.tensor<[12,77,77],f16>, !torch.tensor<[12,77,64],f16> -> !torch.tensor<[12,77,64],f16>
    %int1_609 = torch.constant.int 1
    %int12_610 = torch.constant.int 12
    %int77_611 = torch.constant.int 77
    %int64_612 = torch.constant.int 64
    %854 = torch.prim.ListConstruct %int1_609, %int12_610, %int77_611, %int64_612 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %855 = torch.aten.view %853, %854 : !torch.tensor<[12,77,64],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,64],f16>
    %int1_613 = torch.constant.int 1
    %int2_614 = torch.constant.int 2
    %856 = torch.aten.transpose.int %855, %int1_613, %int2_614 : !torch.tensor<[1,12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int0_615 = torch.constant.int 0
    %857 = torch.aten.clone %856, %int0_615 : !torch.tensor<[1,77,12,64],f16>, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int1_616 = torch.constant.int 1
    %int77_617 = torch.constant.int 77
    %int768_618 = torch.constant.int 768
    %858 = torch.prim.ListConstruct %int1_616, %int77_617, %int768_618 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %859 = torch.aten._unsafe_view %857, %858 : !torch.tensor<[1,77,12,64],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %860 = torch.aten.t %305 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_619 = torch.constant.int 77
    %int768_620 = torch.constant.int 768
    %861 = torch.prim.ListConstruct %int77_619, %int768_620 : (!torch.int, !torch.int) -> !torch.list<int>
    %862 = torch.aten.view %859, %861 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_621 = torch.constant.int 1
    %int1_622 = torch.constant.int 1
    %863 = torch.aten.addmm %304, %862, %860, %int1_621, %int1_622 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_623 = torch.constant.int 1
    %int77_624 = torch.constant.int 77
    %int768_625 = torch.constant.int 768
    %864 = torch.prim.ListConstruct %int1_623, %int77_624, %int768_625 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %865 = torch.aten.view %863, %864 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_626 = torch.constant.int 1
    %866 = torch.aten.add.Tensor %805, %865, %int1_626 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_627 = torch.constant.int 768
    %867 = torch.prim.ListConstruct %int768_627 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_628 = torch.constant.float 1.000000e-05
    %result0_629, %result1_630, %result2_631 = torch.aten.native_layer_norm %866, %867, %303, %302, %float1.000000e-05_628 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %868 = torch.aten.t %301 : !torch.tensor -> !torch.tensor<[768,3072],f16>
    %int77_632 = torch.constant.int 77
    %int768_633 = torch.constant.int 768
    %869 = torch.prim.ListConstruct %int77_632, %int768_633 : (!torch.int, !torch.int) -> !torch.list<int>
    %870 = torch.aten.view %result0_629, %869 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_634 = torch.constant.int 1
    %int1_635 = torch.constant.int 1
    %871 = torch.aten.addmm %300, %870, %868, %int1_634, %int1_635 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,3072],f16>, !torch.int, !torch.int -> !torch.tensor<[77,3072],f16>
    %int1_636 = torch.constant.int 1
    %int77_637 = torch.constant.int 77
    %int3072_638 = torch.constant.int 3072
    %872 = torch.prim.ListConstruct %int1_636, %int77_637, %int3072_638 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %873 = torch.aten.view %871, %872 : !torch.tensor<[77,3072],f16>, !torch.list<int> -> !torch.tensor<[1,77,3072],f16>
    %874 = torch.tensor.literal(dense<1.702000e+00> : tensor<f64>) : !torch.tensor<[],f64>
    %875 = torch.aten.mul.Tensor %873, %874 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,3072],f16>
    %876 = torch.aten.sigmoid %875 : !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %877 = torch.aten.mul.Tensor %873, %876 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %878 = torch.aten.t %299 : !torch.tensor -> !torch.tensor<[3072,768],f16>
    %int77_639 = torch.constant.int 77
    %int3072_640 = torch.constant.int 3072
    %879 = torch.prim.ListConstruct %int77_639, %int3072_640 : (!torch.int, !torch.int) -> !torch.list<int>
    %880 = torch.aten.view %877, %879 : !torch.tensor<[1,77,3072],f16>, !torch.list<int> -> !torch.tensor<[77,3072],f16>
    %int1_641 = torch.constant.int 1
    %int1_642 = torch.constant.int 1
    %881 = torch.aten.addmm %298, %880, %878, %int1_641, %int1_642 : !torch.tensor, !torch.tensor<[77,3072],f16>, !torch.tensor<[3072,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_643 = torch.constant.int 1
    %int77_644 = torch.constant.int 77
    %int768_645 = torch.constant.int 768
    %882 = torch.prim.ListConstruct %int1_643, %int77_644, %int768_645 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %883 = torch.aten.view %881, %882 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_646 = torch.constant.int 1
    %884 = torch.aten.add.Tensor %866, %883, %int1_646 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_647 = torch.constant.int 768
    %885 = torch.prim.ListConstruct %int768_647 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_648 = torch.constant.float 1.000000e-05
    %result0_649, %result1_650, %result2_651 = torch.aten.native_layer_norm %884, %885, %297, %296, %float1.000000e-05_648 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %886 = torch.aten.t %295 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_652 = torch.constant.int 77
    %int768_653 = torch.constant.int 768
    %887 = torch.prim.ListConstruct %int77_652, %int768_653 : (!torch.int, !torch.int) -> !torch.list<int>
    %888 = torch.aten.view %result0_649, %887 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_654 = torch.constant.int 1
    %int1_655 = torch.constant.int 1
    %889 = torch.aten.addmm %294, %888, %886, %int1_654, %int1_655 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_656 = torch.constant.int 1
    %int77_657 = torch.constant.int 77
    %int768_658 = torch.constant.int 768
    %890 = torch.prim.ListConstruct %int1_656, %int77_657, %int768_658 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %891 = torch.aten.view %889, %890 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %892 = torch.tensor.literal(dense<1.250000e-01> : tensor<f64>) : !torch.tensor<[],f64>
    %893 = torch.aten.mul.Tensor %891, %892 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,768],f16>
    %894 = torch.aten.t %293 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_659 = torch.constant.int 77
    %int768_660 = torch.constant.int 768
    %895 = torch.prim.ListConstruct %int77_659, %int768_660 : (!torch.int, !torch.int) -> !torch.list<int>
    %896 = torch.aten.view %result0_649, %895 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_661 = torch.constant.int 1
    %int1_662 = torch.constant.int 1
    %897 = torch.aten.addmm %292, %896, %894, %int1_661, %int1_662 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_663 = torch.constant.int 1
    %int77_664 = torch.constant.int 77
    %int768_665 = torch.constant.int 768
    %898 = torch.prim.ListConstruct %int1_663, %int77_664, %int768_665 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %899 = torch.aten.view %897, %898 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_666 = torch.constant.int 1
    %int-1_667 = torch.constant.int -1
    %int12_668 = torch.constant.int 12
    %int64_669 = torch.constant.int 64
    %900 = torch.prim.ListConstruct %int1_666, %int-1_667, %int12_668, %int64_669 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %901 = torch.aten.view %899, %900 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_670 = torch.constant.int 1
    %int2_671 = torch.constant.int 2
    %902 = torch.aten.transpose.int %901, %int1_670, %int2_671 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_672 = torch.constant.int 0
    %903 = torch.aten.clone %902, %int0_672 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %904 = torch.aten.t %291 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_673 = torch.constant.int 77
    %int768_674 = torch.constant.int 768
    %905 = torch.prim.ListConstruct %int77_673, %int768_674 : (!torch.int, !torch.int) -> !torch.list<int>
    %906 = torch.aten.view %result0_649, %905 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_675 = torch.constant.int 1
    %int1_676 = torch.constant.int 1
    %907 = torch.aten.addmm %290, %906, %904, %int1_675, %int1_676 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_677 = torch.constant.int 1
    %int77_678 = torch.constant.int 77
    %int768_679 = torch.constant.int 768
    %908 = torch.prim.ListConstruct %int1_677, %int77_678, %int768_679 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %909 = torch.aten.view %907, %908 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_680 = torch.constant.int 1
    %int-1_681 = torch.constant.int -1
    %int12_682 = torch.constant.int 12
    %int64_683 = torch.constant.int 64
    %910 = torch.prim.ListConstruct %int1_680, %int-1_681, %int12_682, %int64_683 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %911 = torch.aten.view %909, %910 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_684 = torch.constant.int 1
    %int2_685 = torch.constant.int 2
    %912 = torch.aten.transpose.int %911, %int1_684, %int2_685 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_686 = torch.constant.int 0
    %913 = torch.aten.clone %912, %int0_686 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int1_687 = torch.constant.int 1
    %int77_688 = torch.constant.int 77
    %int12_689 = torch.constant.int 12
    %int64_690 = torch.constant.int 64
    %914 = torch.prim.ListConstruct %int1_687, %int77_688, %int12_689, %int64_690 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %915 = torch.aten.view %893, %914 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_691 = torch.constant.int 1
    %int2_692 = torch.constant.int 2
    %916 = torch.aten.transpose.int %915, %int1_691, %int2_692 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_693 = torch.constant.int 0
    %917 = torch.aten.clone %916, %int0_693 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int12_694 = torch.constant.int 12
    %int-1_695 = torch.constant.int -1
    %int64_696 = torch.constant.int 64
    %918 = torch.prim.ListConstruct %int12_694, %int-1_695, %int64_696 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %919 = torch.aten.view %917, %918 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_697 = torch.constant.int 12
    %int-1_698 = torch.constant.int -1
    %int64_699 = torch.constant.int 64
    %920 = torch.prim.ListConstruct %int12_697, %int-1_698, %int64_699 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %921 = torch.aten.view %903, %920 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_700 = torch.constant.int 12
    %int-1_701 = torch.constant.int -1
    %int64_702 = torch.constant.int 64
    %922 = torch.prim.ListConstruct %int12_700, %int-1_701, %int64_702 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %923 = torch.aten.view %913, %922 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int1_703 = torch.constant.int 1
    %int2_704 = torch.constant.int 2
    %924 = torch.aten.transpose.int %921, %int1_703, %int2_704 : !torch.tensor<[12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[12,64,77],f16>
    %925 = torch.aten.bmm %919, %924 : !torch.tensor<[12,77,64],f16>, !torch.tensor<[12,64,77],f16> -> !torch.tensor<[12,77,77],f16>
    %int1_705 = torch.constant.int 1
    %int12_706 = torch.constant.int 12
    %int77_707 = torch.constant.int 77
    %int77_708 = torch.constant.int 77
    %926 = torch.prim.ListConstruct %int1_705, %int12_706, %int77_707, %int77_708 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %927 = torch.aten.view %925, %926 : !torch.tensor<[12,77,77],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,77],f16>
    %int1_709 = torch.constant.int 1
    %928 = torch.aten.add.Tensor %927, %410, %int1_709 : !torch.tensor<[1,12,77,77],f16>, !torch.tensor<[1,1,77,77],f16>, !torch.int -> !torch.tensor<[1,12,77,77],f16>
    %int12_710 = torch.constant.int 12
    %int77_711 = torch.constant.int 77
    %int77_712 = torch.constant.int 77
    %929 = torch.prim.ListConstruct %int12_710, %int77_711, %int77_712 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %930 = torch.aten.view %928, %929 : !torch.tensor<[1,12,77,77],f16>, !torch.list<int> -> !torch.tensor<[12,77,77],f16>
    %int-1_713 = torch.constant.int -1
    %false_714 = torch.constant.bool false
    %931 = torch.aten._softmax %930, %int-1_713, %false_714 : !torch.tensor<[12,77,77],f16>, !torch.int, !torch.bool -> !torch.tensor<[12,77,77],f16>
    %932 = torch.aten.bmm %931, %923 : !torch.tensor<[12,77,77],f16>, !torch.tensor<[12,77,64],f16> -> !torch.tensor<[12,77,64],f16>
    %int1_715 = torch.constant.int 1
    %int12_716 = torch.constant.int 12
    %int77_717 = torch.constant.int 77
    %int64_718 = torch.constant.int 64
    %933 = torch.prim.ListConstruct %int1_715, %int12_716, %int77_717, %int64_718 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %934 = torch.aten.view %932, %933 : !torch.tensor<[12,77,64],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,64],f16>
    %int1_719 = torch.constant.int 1
    %int2_720 = torch.constant.int 2
    %935 = torch.aten.transpose.int %934, %int1_719, %int2_720 : !torch.tensor<[1,12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int0_721 = torch.constant.int 0
    %936 = torch.aten.clone %935, %int0_721 : !torch.tensor<[1,77,12,64],f16>, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int1_722 = torch.constant.int 1
    %int77_723 = torch.constant.int 77
    %int768_724 = torch.constant.int 768
    %937 = torch.prim.ListConstruct %int1_722, %int77_723, %int768_724 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %938 = torch.aten._unsafe_view %936, %937 : !torch.tensor<[1,77,12,64],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %939 = torch.aten.t %289 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_725 = torch.constant.int 77
    %int768_726 = torch.constant.int 768
    %940 = torch.prim.ListConstruct %int77_725, %int768_726 : (!torch.int, !torch.int) -> !torch.list<int>
    %941 = torch.aten.view %938, %940 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_727 = torch.constant.int 1
    %int1_728 = torch.constant.int 1
    %942 = torch.aten.addmm %288, %941, %939, %int1_727, %int1_728 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_729 = torch.constant.int 1
    %int77_730 = torch.constant.int 77
    %int768_731 = torch.constant.int 768
    %943 = torch.prim.ListConstruct %int1_729, %int77_730, %int768_731 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %944 = torch.aten.view %942, %943 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_732 = torch.constant.int 1
    %945 = torch.aten.add.Tensor %884, %944, %int1_732 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_733 = torch.constant.int 768
    %946 = torch.prim.ListConstruct %int768_733 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_734 = torch.constant.float 1.000000e-05
    %result0_735, %result1_736, %result2_737 = torch.aten.native_layer_norm %945, %946, %287, %286, %float1.000000e-05_734 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %947 = torch.aten.t %285 : !torch.tensor -> !torch.tensor<[768,3072],f16>
    %int77_738 = torch.constant.int 77
    %int768_739 = torch.constant.int 768
    %948 = torch.prim.ListConstruct %int77_738, %int768_739 : (!torch.int, !torch.int) -> !torch.list<int>
    %949 = torch.aten.view %result0_735, %948 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_740 = torch.constant.int 1
    %int1_741 = torch.constant.int 1
    %950 = torch.aten.addmm %284, %949, %947, %int1_740, %int1_741 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,3072],f16>, !torch.int, !torch.int -> !torch.tensor<[77,3072],f16>
    %int1_742 = torch.constant.int 1
    %int77_743 = torch.constant.int 77
    %int3072_744 = torch.constant.int 3072
    %951 = torch.prim.ListConstruct %int1_742, %int77_743, %int3072_744 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %952 = torch.aten.view %950, %951 : !torch.tensor<[77,3072],f16>, !torch.list<int> -> !torch.tensor<[1,77,3072],f16>
    %953 = torch.tensor.literal(dense<1.702000e+00> : tensor<f64>) : !torch.tensor<[],f64>
    %954 = torch.aten.mul.Tensor %952, %953 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,3072],f16>
    %955 = torch.aten.sigmoid %954 : !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %956 = torch.aten.mul.Tensor %952, %955 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %957 = torch.aten.t %283 : !torch.tensor -> !torch.tensor<[3072,768],f16>
    %int77_745 = torch.constant.int 77
    %int3072_746 = torch.constant.int 3072
    %958 = torch.prim.ListConstruct %int77_745, %int3072_746 : (!torch.int, !torch.int) -> !torch.list<int>
    %959 = torch.aten.view %956, %958 : !torch.tensor<[1,77,3072],f16>, !torch.list<int> -> !torch.tensor<[77,3072],f16>
    %int1_747 = torch.constant.int 1
    %int1_748 = torch.constant.int 1
    %960 = torch.aten.addmm %282, %959, %957, %int1_747, %int1_748 : !torch.tensor, !torch.tensor<[77,3072],f16>, !torch.tensor<[3072,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_749 = torch.constant.int 1
    %int77_750 = torch.constant.int 77
    %int768_751 = torch.constant.int 768
    %961 = torch.prim.ListConstruct %int1_749, %int77_750, %int768_751 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %962 = torch.aten.view %960, %961 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_752 = torch.constant.int 1
    %963 = torch.aten.add.Tensor %945, %962, %int1_752 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_753 = torch.constant.int 768
    %964 = torch.prim.ListConstruct %int768_753 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_754 = torch.constant.float 1.000000e-05
    %result0_755, %result1_756, %result2_757 = torch.aten.native_layer_norm %963, %964, %281, %280, %float1.000000e-05_754 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %965 = torch.aten.t %279 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_758 = torch.constant.int 77
    %int768_759 = torch.constant.int 768
    %966 = torch.prim.ListConstruct %int77_758, %int768_759 : (!torch.int, !torch.int) -> !torch.list<int>
    %967 = torch.aten.view %result0_755, %966 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_760 = torch.constant.int 1
    %int1_761 = torch.constant.int 1
    %968 = torch.aten.addmm %278, %967, %965, %int1_760, %int1_761 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_762 = torch.constant.int 1
    %int77_763 = torch.constant.int 77
    %int768_764 = torch.constant.int 768
    %969 = torch.prim.ListConstruct %int1_762, %int77_763, %int768_764 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %970 = torch.aten.view %968, %969 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %971 = torch.tensor.literal(dense<1.250000e-01> : tensor<f64>) : !torch.tensor<[],f64>
    %972 = torch.aten.mul.Tensor %970, %971 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,768],f16>
    %973 = torch.aten.t %277 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_765 = torch.constant.int 77
    %int768_766 = torch.constant.int 768
    %974 = torch.prim.ListConstruct %int77_765, %int768_766 : (!torch.int, !torch.int) -> !torch.list<int>
    %975 = torch.aten.view %result0_755, %974 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_767 = torch.constant.int 1
    %int1_768 = torch.constant.int 1
    %976 = torch.aten.addmm %276, %975, %973, %int1_767, %int1_768 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_769 = torch.constant.int 1
    %int77_770 = torch.constant.int 77
    %int768_771 = torch.constant.int 768
    %977 = torch.prim.ListConstruct %int1_769, %int77_770, %int768_771 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %978 = torch.aten.view %976, %977 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_772 = torch.constant.int 1
    %int-1_773 = torch.constant.int -1
    %int12_774 = torch.constant.int 12
    %int64_775 = torch.constant.int 64
    %979 = torch.prim.ListConstruct %int1_772, %int-1_773, %int12_774, %int64_775 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %980 = torch.aten.view %978, %979 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_776 = torch.constant.int 1
    %int2_777 = torch.constant.int 2
    %981 = torch.aten.transpose.int %980, %int1_776, %int2_777 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_778 = torch.constant.int 0
    %982 = torch.aten.clone %981, %int0_778 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %983 = torch.aten.t %275 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_779 = torch.constant.int 77
    %int768_780 = torch.constant.int 768
    %984 = torch.prim.ListConstruct %int77_779, %int768_780 : (!torch.int, !torch.int) -> !torch.list<int>
    %985 = torch.aten.view %result0_755, %984 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_781 = torch.constant.int 1
    %int1_782 = torch.constant.int 1
    %986 = torch.aten.addmm %274, %985, %983, %int1_781, %int1_782 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_783 = torch.constant.int 1
    %int77_784 = torch.constant.int 77
    %int768_785 = torch.constant.int 768
    %987 = torch.prim.ListConstruct %int1_783, %int77_784, %int768_785 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %988 = torch.aten.view %986, %987 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_786 = torch.constant.int 1
    %int-1_787 = torch.constant.int -1
    %int12_788 = torch.constant.int 12
    %int64_789 = torch.constant.int 64
    %989 = torch.prim.ListConstruct %int1_786, %int-1_787, %int12_788, %int64_789 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %990 = torch.aten.view %988, %989 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_790 = torch.constant.int 1
    %int2_791 = torch.constant.int 2
    %991 = torch.aten.transpose.int %990, %int1_790, %int2_791 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_792 = torch.constant.int 0
    %992 = torch.aten.clone %991, %int0_792 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int1_793 = torch.constant.int 1
    %int77_794 = torch.constant.int 77
    %int12_795 = torch.constant.int 12
    %int64_796 = torch.constant.int 64
    %993 = torch.prim.ListConstruct %int1_793, %int77_794, %int12_795, %int64_796 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %994 = torch.aten.view %972, %993 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_797 = torch.constant.int 1
    %int2_798 = torch.constant.int 2
    %995 = torch.aten.transpose.int %994, %int1_797, %int2_798 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_799 = torch.constant.int 0
    %996 = torch.aten.clone %995, %int0_799 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int12_800 = torch.constant.int 12
    %int-1_801 = torch.constant.int -1
    %int64_802 = torch.constant.int 64
    %997 = torch.prim.ListConstruct %int12_800, %int-1_801, %int64_802 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %998 = torch.aten.view %996, %997 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_803 = torch.constant.int 12
    %int-1_804 = torch.constant.int -1
    %int64_805 = torch.constant.int 64
    %999 = torch.prim.ListConstruct %int12_803, %int-1_804, %int64_805 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1000 = torch.aten.view %982, %999 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_806 = torch.constant.int 12
    %int-1_807 = torch.constant.int -1
    %int64_808 = torch.constant.int 64
    %1001 = torch.prim.ListConstruct %int12_806, %int-1_807, %int64_808 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1002 = torch.aten.view %992, %1001 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int1_809 = torch.constant.int 1
    %int2_810 = torch.constant.int 2
    %1003 = torch.aten.transpose.int %1000, %int1_809, %int2_810 : !torch.tensor<[12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[12,64,77],f16>
    %1004 = torch.aten.bmm %998, %1003 : !torch.tensor<[12,77,64],f16>, !torch.tensor<[12,64,77],f16> -> !torch.tensor<[12,77,77],f16>
    %int1_811 = torch.constant.int 1
    %int12_812 = torch.constant.int 12
    %int77_813 = torch.constant.int 77
    %int77_814 = torch.constant.int 77
    %1005 = torch.prim.ListConstruct %int1_811, %int12_812, %int77_813, %int77_814 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1006 = torch.aten.view %1004, %1005 : !torch.tensor<[12,77,77],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,77],f16>
    %int1_815 = torch.constant.int 1
    %1007 = torch.aten.add.Tensor %1006, %410, %int1_815 : !torch.tensor<[1,12,77,77],f16>, !torch.tensor<[1,1,77,77],f16>, !torch.int -> !torch.tensor<[1,12,77,77],f16>
    %int12_816 = torch.constant.int 12
    %int77_817 = torch.constant.int 77
    %int77_818 = torch.constant.int 77
    %1008 = torch.prim.ListConstruct %int12_816, %int77_817, %int77_818 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1009 = torch.aten.view %1007, %1008 : !torch.tensor<[1,12,77,77],f16>, !torch.list<int> -> !torch.tensor<[12,77,77],f16>
    %int-1_819 = torch.constant.int -1
    %false_820 = torch.constant.bool false
    %1010 = torch.aten._softmax %1009, %int-1_819, %false_820 : !torch.tensor<[12,77,77],f16>, !torch.int, !torch.bool -> !torch.tensor<[12,77,77],f16>
    %1011 = torch.aten.bmm %1010, %1002 : !torch.tensor<[12,77,77],f16>, !torch.tensor<[12,77,64],f16> -> !torch.tensor<[12,77,64],f16>
    %int1_821 = torch.constant.int 1
    %int12_822 = torch.constant.int 12
    %int77_823 = torch.constant.int 77
    %int64_824 = torch.constant.int 64
    %1012 = torch.prim.ListConstruct %int1_821, %int12_822, %int77_823, %int64_824 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1013 = torch.aten.view %1011, %1012 : !torch.tensor<[12,77,64],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,64],f16>
    %int1_825 = torch.constant.int 1
    %int2_826 = torch.constant.int 2
    %1014 = torch.aten.transpose.int %1013, %int1_825, %int2_826 : !torch.tensor<[1,12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int0_827 = torch.constant.int 0
    %1015 = torch.aten.clone %1014, %int0_827 : !torch.tensor<[1,77,12,64],f16>, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int1_828 = torch.constant.int 1
    %int77_829 = torch.constant.int 77
    %int768_830 = torch.constant.int 768
    %1016 = torch.prim.ListConstruct %int1_828, %int77_829, %int768_830 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1017 = torch.aten._unsafe_view %1015, %1016 : !torch.tensor<[1,77,12,64],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %1018 = torch.aten.t %273 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_831 = torch.constant.int 77
    %int768_832 = torch.constant.int 768
    %1019 = torch.prim.ListConstruct %int77_831, %int768_832 : (!torch.int, !torch.int) -> !torch.list<int>
    %1020 = torch.aten.view %1017, %1019 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_833 = torch.constant.int 1
    %int1_834 = torch.constant.int 1
    %1021 = torch.aten.addmm %272, %1020, %1018, %int1_833, %int1_834 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_835 = torch.constant.int 1
    %int77_836 = torch.constant.int 77
    %int768_837 = torch.constant.int 768
    %1022 = torch.prim.ListConstruct %int1_835, %int77_836, %int768_837 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1023 = torch.aten.view %1021, %1022 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_838 = torch.constant.int 1
    %1024 = torch.aten.add.Tensor %963, %1023, %int1_838 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_839 = torch.constant.int 768
    %1025 = torch.prim.ListConstruct %int768_839 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_840 = torch.constant.float 1.000000e-05
    %result0_841, %result1_842, %result2_843 = torch.aten.native_layer_norm %1024, %1025, %271, %270, %float1.000000e-05_840 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %1026 = torch.aten.t %269 : !torch.tensor -> !torch.tensor<[768,3072],f16>
    %int77_844 = torch.constant.int 77
    %int768_845 = torch.constant.int 768
    %1027 = torch.prim.ListConstruct %int77_844, %int768_845 : (!torch.int, !torch.int) -> !torch.list<int>
    %1028 = torch.aten.view %result0_841, %1027 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_846 = torch.constant.int 1
    %int1_847 = torch.constant.int 1
    %1029 = torch.aten.addmm %268, %1028, %1026, %int1_846, %int1_847 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,3072],f16>, !torch.int, !torch.int -> !torch.tensor<[77,3072],f16>
    %int1_848 = torch.constant.int 1
    %int77_849 = torch.constant.int 77
    %int3072_850 = torch.constant.int 3072
    %1030 = torch.prim.ListConstruct %int1_848, %int77_849, %int3072_850 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1031 = torch.aten.view %1029, %1030 : !torch.tensor<[77,3072],f16>, !torch.list<int> -> !torch.tensor<[1,77,3072],f16>
    %1032 = torch.tensor.literal(dense<1.702000e+00> : tensor<f64>) : !torch.tensor<[],f64>
    %1033 = torch.aten.mul.Tensor %1031, %1032 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,3072],f16>
    %1034 = torch.aten.sigmoid %1033 : !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %1035 = torch.aten.mul.Tensor %1031, %1034 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %1036 = torch.aten.t %267 : !torch.tensor -> !torch.tensor<[3072,768],f16>
    %int77_851 = torch.constant.int 77
    %int3072_852 = torch.constant.int 3072
    %1037 = torch.prim.ListConstruct %int77_851, %int3072_852 : (!torch.int, !torch.int) -> !torch.list<int>
    %1038 = torch.aten.view %1035, %1037 : !torch.tensor<[1,77,3072],f16>, !torch.list<int> -> !torch.tensor<[77,3072],f16>
    %int1_853 = torch.constant.int 1
    %int1_854 = torch.constant.int 1
    %1039 = torch.aten.addmm %266, %1038, %1036, %int1_853, %int1_854 : !torch.tensor, !torch.tensor<[77,3072],f16>, !torch.tensor<[3072,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_855 = torch.constant.int 1
    %int77_856 = torch.constant.int 77
    %int768_857 = torch.constant.int 768
    %1040 = torch.prim.ListConstruct %int1_855, %int77_856, %int768_857 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1041 = torch.aten.view %1039, %1040 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_858 = torch.constant.int 1
    %1042 = torch.aten.add.Tensor %1024, %1041, %int1_858 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_859 = torch.constant.int 768
    %1043 = torch.prim.ListConstruct %int768_859 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_860 = torch.constant.float 1.000000e-05
    %result0_861, %result1_862, %result2_863 = torch.aten.native_layer_norm %1042, %1043, %265, %264, %float1.000000e-05_860 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %1044 = torch.aten.t %263 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_864 = torch.constant.int 77
    %int768_865 = torch.constant.int 768
    %1045 = torch.prim.ListConstruct %int77_864, %int768_865 : (!torch.int, !torch.int) -> !torch.list<int>
    %1046 = torch.aten.view %result0_861, %1045 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_866 = torch.constant.int 1
    %int1_867 = torch.constant.int 1
    %1047 = torch.aten.addmm %262, %1046, %1044, %int1_866, %int1_867 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_868 = torch.constant.int 1
    %int77_869 = torch.constant.int 77
    %int768_870 = torch.constant.int 768
    %1048 = torch.prim.ListConstruct %int1_868, %int77_869, %int768_870 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1049 = torch.aten.view %1047, %1048 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %1050 = torch.tensor.literal(dense<1.250000e-01> : tensor<f64>) : !torch.tensor<[],f64>
    %1051 = torch.aten.mul.Tensor %1049, %1050 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,768],f16>
    %1052 = torch.aten.t %261 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_871 = torch.constant.int 77
    %int768_872 = torch.constant.int 768
    %1053 = torch.prim.ListConstruct %int77_871, %int768_872 : (!torch.int, !torch.int) -> !torch.list<int>
    %1054 = torch.aten.view %result0_861, %1053 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_873 = torch.constant.int 1
    %int1_874 = torch.constant.int 1
    %1055 = torch.aten.addmm %260, %1054, %1052, %int1_873, %int1_874 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_875 = torch.constant.int 1
    %int77_876 = torch.constant.int 77
    %int768_877 = torch.constant.int 768
    %1056 = torch.prim.ListConstruct %int1_875, %int77_876, %int768_877 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1057 = torch.aten.view %1055, %1056 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_878 = torch.constant.int 1
    %int-1_879 = torch.constant.int -1
    %int12_880 = torch.constant.int 12
    %int64_881 = torch.constant.int 64
    %1058 = torch.prim.ListConstruct %int1_878, %int-1_879, %int12_880, %int64_881 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1059 = torch.aten.view %1057, %1058 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_882 = torch.constant.int 1
    %int2_883 = torch.constant.int 2
    %1060 = torch.aten.transpose.int %1059, %int1_882, %int2_883 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_884 = torch.constant.int 0
    %1061 = torch.aten.clone %1060, %int0_884 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %1062 = torch.aten.t %259 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_885 = torch.constant.int 77
    %int768_886 = torch.constant.int 768
    %1063 = torch.prim.ListConstruct %int77_885, %int768_886 : (!torch.int, !torch.int) -> !torch.list<int>
    %1064 = torch.aten.view %result0_861, %1063 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_887 = torch.constant.int 1
    %int1_888 = torch.constant.int 1
    %1065 = torch.aten.addmm %258, %1064, %1062, %int1_887, %int1_888 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_889 = torch.constant.int 1
    %int77_890 = torch.constant.int 77
    %int768_891 = torch.constant.int 768
    %1066 = torch.prim.ListConstruct %int1_889, %int77_890, %int768_891 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1067 = torch.aten.view %1065, %1066 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_892 = torch.constant.int 1
    %int-1_893 = torch.constant.int -1
    %int12_894 = torch.constant.int 12
    %int64_895 = torch.constant.int 64
    %1068 = torch.prim.ListConstruct %int1_892, %int-1_893, %int12_894, %int64_895 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1069 = torch.aten.view %1067, %1068 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_896 = torch.constant.int 1
    %int2_897 = torch.constant.int 2
    %1070 = torch.aten.transpose.int %1069, %int1_896, %int2_897 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_898 = torch.constant.int 0
    %1071 = torch.aten.clone %1070, %int0_898 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int1_899 = torch.constant.int 1
    %int77_900 = torch.constant.int 77
    %int12_901 = torch.constant.int 12
    %int64_902 = torch.constant.int 64
    %1072 = torch.prim.ListConstruct %int1_899, %int77_900, %int12_901, %int64_902 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1073 = torch.aten.view %1051, %1072 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_903 = torch.constant.int 1
    %int2_904 = torch.constant.int 2
    %1074 = torch.aten.transpose.int %1073, %int1_903, %int2_904 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_905 = torch.constant.int 0
    %1075 = torch.aten.clone %1074, %int0_905 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int12_906 = torch.constant.int 12
    %int-1_907 = torch.constant.int -1
    %int64_908 = torch.constant.int 64
    %1076 = torch.prim.ListConstruct %int12_906, %int-1_907, %int64_908 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1077 = torch.aten.view %1075, %1076 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_909 = torch.constant.int 12
    %int-1_910 = torch.constant.int -1
    %int64_911 = torch.constant.int 64
    %1078 = torch.prim.ListConstruct %int12_909, %int-1_910, %int64_911 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1079 = torch.aten.view %1061, %1078 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_912 = torch.constant.int 12
    %int-1_913 = torch.constant.int -1
    %int64_914 = torch.constant.int 64
    %1080 = torch.prim.ListConstruct %int12_912, %int-1_913, %int64_914 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1081 = torch.aten.view %1071, %1080 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int1_915 = torch.constant.int 1
    %int2_916 = torch.constant.int 2
    %1082 = torch.aten.transpose.int %1079, %int1_915, %int2_916 : !torch.tensor<[12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[12,64,77],f16>
    %1083 = torch.aten.bmm %1077, %1082 : !torch.tensor<[12,77,64],f16>, !torch.tensor<[12,64,77],f16> -> !torch.tensor<[12,77,77],f16>
    %int1_917 = torch.constant.int 1
    %int12_918 = torch.constant.int 12
    %int77_919 = torch.constant.int 77
    %int77_920 = torch.constant.int 77
    %1084 = torch.prim.ListConstruct %int1_917, %int12_918, %int77_919, %int77_920 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1085 = torch.aten.view %1083, %1084 : !torch.tensor<[12,77,77],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,77],f16>
    %int1_921 = torch.constant.int 1
    %1086 = torch.aten.add.Tensor %1085, %410, %int1_921 : !torch.tensor<[1,12,77,77],f16>, !torch.tensor<[1,1,77,77],f16>, !torch.int -> !torch.tensor<[1,12,77,77],f16>
    %int12_922 = torch.constant.int 12
    %int77_923 = torch.constant.int 77
    %int77_924 = torch.constant.int 77
    %1087 = torch.prim.ListConstruct %int12_922, %int77_923, %int77_924 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1088 = torch.aten.view %1086, %1087 : !torch.tensor<[1,12,77,77],f16>, !torch.list<int> -> !torch.tensor<[12,77,77],f16>
    %int-1_925 = torch.constant.int -1
    %false_926 = torch.constant.bool false
    %1089 = torch.aten._softmax %1088, %int-1_925, %false_926 : !torch.tensor<[12,77,77],f16>, !torch.int, !torch.bool -> !torch.tensor<[12,77,77],f16>
    %1090 = torch.aten.bmm %1089, %1081 : !torch.tensor<[12,77,77],f16>, !torch.tensor<[12,77,64],f16> -> !torch.tensor<[12,77,64],f16>
    %int1_927 = torch.constant.int 1
    %int12_928 = torch.constant.int 12
    %int77_929 = torch.constant.int 77
    %int64_930 = torch.constant.int 64
    %1091 = torch.prim.ListConstruct %int1_927, %int12_928, %int77_929, %int64_930 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1092 = torch.aten.view %1090, %1091 : !torch.tensor<[12,77,64],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,64],f16>
    %int1_931 = torch.constant.int 1
    %int2_932 = torch.constant.int 2
    %1093 = torch.aten.transpose.int %1092, %int1_931, %int2_932 : !torch.tensor<[1,12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int0_933 = torch.constant.int 0
    %1094 = torch.aten.clone %1093, %int0_933 : !torch.tensor<[1,77,12,64],f16>, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int1_934 = torch.constant.int 1
    %int77_935 = torch.constant.int 77
    %int768_936 = torch.constant.int 768
    %1095 = torch.prim.ListConstruct %int1_934, %int77_935, %int768_936 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1096 = torch.aten._unsafe_view %1094, %1095 : !torch.tensor<[1,77,12,64],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %1097 = torch.aten.t %257 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_937 = torch.constant.int 77
    %int768_938 = torch.constant.int 768
    %1098 = torch.prim.ListConstruct %int77_937, %int768_938 : (!torch.int, !torch.int) -> !torch.list<int>
    %1099 = torch.aten.view %1096, %1098 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_939 = torch.constant.int 1
    %int1_940 = torch.constant.int 1
    %1100 = torch.aten.addmm %256, %1099, %1097, %int1_939, %int1_940 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_941 = torch.constant.int 1
    %int77_942 = torch.constant.int 77
    %int768_943 = torch.constant.int 768
    %1101 = torch.prim.ListConstruct %int1_941, %int77_942, %int768_943 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1102 = torch.aten.view %1100, %1101 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_944 = torch.constant.int 1
    %1103 = torch.aten.add.Tensor %1042, %1102, %int1_944 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_945 = torch.constant.int 768
    %1104 = torch.prim.ListConstruct %int768_945 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_946 = torch.constant.float 1.000000e-05
    %result0_947, %result1_948, %result2_949 = torch.aten.native_layer_norm %1103, %1104, %255, %254, %float1.000000e-05_946 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %1105 = torch.aten.t %253 : !torch.tensor -> !torch.tensor<[768,3072],f16>
    %int77_950 = torch.constant.int 77
    %int768_951 = torch.constant.int 768
    %1106 = torch.prim.ListConstruct %int77_950, %int768_951 : (!torch.int, !torch.int) -> !torch.list<int>
    %1107 = torch.aten.view %result0_947, %1106 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_952 = torch.constant.int 1
    %int1_953 = torch.constant.int 1
    %1108 = torch.aten.addmm %252, %1107, %1105, %int1_952, %int1_953 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,3072],f16>, !torch.int, !torch.int -> !torch.tensor<[77,3072],f16>
    %int1_954 = torch.constant.int 1
    %int77_955 = torch.constant.int 77
    %int3072_956 = torch.constant.int 3072
    %1109 = torch.prim.ListConstruct %int1_954, %int77_955, %int3072_956 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1110 = torch.aten.view %1108, %1109 : !torch.tensor<[77,3072],f16>, !torch.list<int> -> !torch.tensor<[1,77,3072],f16>
    %1111 = torch.tensor.literal(dense<1.702000e+00> : tensor<f64>) : !torch.tensor<[],f64>
    %1112 = torch.aten.mul.Tensor %1110, %1111 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,3072],f16>
    %1113 = torch.aten.sigmoid %1112 : !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %1114 = torch.aten.mul.Tensor %1110, %1113 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %1115 = torch.aten.t %251 : !torch.tensor -> !torch.tensor<[3072,768],f16>
    %int77_957 = torch.constant.int 77
    %int3072_958 = torch.constant.int 3072
    %1116 = torch.prim.ListConstruct %int77_957, %int3072_958 : (!torch.int, !torch.int) -> !torch.list<int>
    %1117 = torch.aten.view %1114, %1116 : !torch.tensor<[1,77,3072],f16>, !torch.list<int> -> !torch.tensor<[77,3072],f16>
    %int1_959 = torch.constant.int 1
    %int1_960 = torch.constant.int 1
    %1118 = torch.aten.addmm %250, %1117, %1115, %int1_959, %int1_960 : !torch.tensor, !torch.tensor<[77,3072],f16>, !torch.tensor<[3072,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_961 = torch.constant.int 1
    %int77_962 = torch.constant.int 77
    %int768_963 = torch.constant.int 768
    %1119 = torch.prim.ListConstruct %int1_961, %int77_962, %int768_963 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1120 = torch.aten.view %1118, %1119 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_964 = torch.constant.int 1
    %1121 = torch.aten.add.Tensor %1103, %1120, %int1_964 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_965 = torch.constant.int 768
    %1122 = torch.prim.ListConstruct %int768_965 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_966 = torch.constant.float 1.000000e-05
    %result0_967, %result1_968, %result2_969 = torch.aten.native_layer_norm %1121, %1122, %249, %248, %float1.000000e-05_966 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %1123 = torch.aten.t %247 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_970 = torch.constant.int 77
    %int768_971 = torch.constant.int 768
    %1124 = torch.prim.ListConstruct %int77_970, %int768_971 : (!torch.int, !torch.int) -> !torch.list<int>
    %1125 = torch.aten.view %result0_967, %1124 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_972 = torch.constant.int 1
    %int1_973 = torch.constant.int 1
    %1126 = torch.aten.addmm %246, %1125, %1123, %int1_972, %int1_973 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_974 = torch.constant.int 1
    %int77_975 = torch.constant.int 77
    %int768_976 = torch.constant.int 768
    %1127 = torch.prim.ListConstruct %int1_974, %int77_975, %int768_976 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1128 = torch.aten.view %1126, %1127 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %1129 = torch.tensor.literal(dense<1.250000e-01> : tensor<f64>) : !torch.tensor<[],f64>
    %1130 = torch.aten.mul.Tensor %1128, %1129 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,768],f16>
    %1131 = torch.aten.t %245 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_977 = torch.constant.int 77
    %int768_978 = torch.constant.int 768
    %1132 = torch.prim.ListConstruct %int77_977, %int768_978 : (!torch.int, !torch.int) -> !torch.list<int>
    %1133 = torch.aten.view %result0_967, %1132 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_979 = torch.constant.int 1
    %int1_980 = torch.constant.int 1
    %1134 = torch.aten.addmm %244, %1133, %1131, %int1_979, %int1_980 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_981 = torch.constant.int 1
    %int77_982 = torch.constant.int 77
    %int768_983 = torch.constant.int 768
    %1135 = torch.prim.ListConstruct %int1_981, %int77_982, %int768_983 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1136 = torch.aten.view %1134, %1135 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_984 = torch.constant.int 1
    %int-1_985 = torch.constant.int -1
    %int12_986 = torch.constant.int 12
    %int64_987 = torch.constant.int 64
    %1137 = torch.prim.ListConstruct %int1_984, %int-1_985, %int12_986, %int64_987 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1138 = torch.aten.view %1136, %1137 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_988 = torch.constant.int 1
    %int2_989 = torch.constant.int 2
    %1139 = torch.aten.transpose.int %1138, %int1_988, %int2_989 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_990 = torch.constant.int 0
    %1140 = torch.aten.clone %1139, %int0_990 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %1141 = torch.aten.t %243 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_991 = torch.constant.int 77
    %int768_992 = torch.constant.int 768
    %1142 = torch.prim.ListConstruct %int77_991, %int768_992 : (!torch.int, !torch.int) -> !torch.list<int>
    %1143 = torch.aten.view %result0_967, %1142 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_993 = torch.constant.int 1
    %int1_994 = torch.constant.int 1
    %1144 = torch.aten.addmm %242, %1143, %1141, %int1_993, %int1_994 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_995 = torch.constant.int 1
    %int77_996 = torch.constant.int 77
    %int768_997 = torch.constant.int 768
    %1145 = torch.prim.ListConstruct %int1_995, %int77_996, %int768_997 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1146 = torch.aten.view %1144, %1145 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_998 = torch.constant.int 1
    %int-1_999 = torch.constant.int -1
    %int12_1000 = torch.constant.int 12
    %int64_1001 = torch.constant.int 64
    %1147 = torch.prim.ListConstruct %int1_998, %int-1_999, %int12_1000, %int64_1001 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1148 = torch.aten.view %1146, %1147 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_1002 = torch.constant.int 1
    %int2_1003 = torch.constant.int 2
    %1149 = torch.aten.transpose.int %1148, %int1_1002, %int2_1003 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_1004 = torch.constant.int 0
    %1150 = torch.aten.clone %1149, %int0_1004 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int1_1005 = torch.constant.int 1
    %int77_1006 = torch.constant.int 77
    %int12_1007 = torch.constant.int 12
    %int64_1008 = torch.constant.int 64
    %1151 = torch.prim.ListConstruct %int1_1005, %int77_1006, %int12_1007, %int64_1008 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1152 = torch.aten.view %1130, %1151 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_1009 = torch.constant.int 1
    %int2_1010 = torch.constant.int 2
    %1153 = torch.aten.transpose.int %1152, %int1_1009, %int2_1010 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_1011 = torch.constant.int 0
    %1154 = torch.aten.clone %1153, %int0_1011 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int12_1012 = torch.constant.int 12
    %int-1_1013 = torch.constant.int -1
    %int64_1014 = torch.constant.int 64
    %1155 = torch.prim.ListConstruct %int12_1012, %int-1_1013, %int64_1014 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1156 = torch.aten.view %1154, %1155 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_1015 = torch.constant.int 12
    %int-1_1016 = torch.constant.int -1
    %int64_1017 = torch.constant.int 64
    %1157 = torch.prim.ListConstruct %int12_1015, %int-1_1016, %int64_1017 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1158 = torch.aten.view %1140, %1157 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_1018 = torch.constant.int 12
    %int-1_1019 = torch.constant.int -1
    %int64_1020 = torch.constant.int 64
    %1159 = torch.prim.ListConstruct %int12_1018, %int-1_1019, %int64_1020 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1160 = torch.aten.view %1150, %1159 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int1_1021 = torch.constant.int 1
    %int2_1022 = torch.constant.int 2
    %1161 = torch.aten.transpose.int %1158, %int1_1021, %int2_1022 : !torch.tensor<[12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[12,64,77],f16>
    %1162 = torch.aten.bmm %1156, %1161 : !torch.tensor<[12,77,64],f16>, !torch.tensor<[12,64,77],f16> -> !torch.tensor<[12,77,77],f16>
    %int1_1023 = torch.constant.int 1
    %int12_1024 = torch.constant.int 12
    %int77_1025 = torch.constant.int 77
    %int77_1026 = torch.constant.int 77
    %1163 = torch.prim.ListConstruct %int1_1023, %int12_1024, %int77_1025, %int77_1026 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1164 = torch.aten.view %1162, %1163 : !torch.tensor<[12,77,77],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,77],f16>
    %int1_1027 = torch.constant.int 1
    %1165 = torch.aten.add.Tensor %1164, %410, %int1_1027 : !torch.tensor<[1,12,77,77],f16>, !torch.tensor<[1,1,77,77],f16>, !torch.int -> !torch.tensor<[1,12,77,77],f16>
    %int12_1028 = torch.constant.int 12
    %int77_1029 = torch.constant.int 77
    %int77_1030 = torch.constant.int 77
    %1166 = torch.prim.ListConstruct %int12_1028, %int77_1029, %int77_1030 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1167 = torch.aten.view %1165, %1166 : !torch.tensor<[1,12,77,77],f16>, !torch.list<int> -> !torch.tensor<[12,77,77],f16>
    %int-1_1031 = torch.constant.int -1
    %false_1032 = torch.constant.bool false
    %1168 = torch.aten._softmax %1167, %int-1_1031, %false_1032 : !torch.tensor<[12,77,77],f16>, !torch.int, !torch.bool -> !torch.tensor<[12,77,77],f16>
    %1169 = torch.aten.bmm %1168, %1160 : !torch.tensor<[12,77,77],f16>, !torch.tensor<[12,77,64],f16> -> !torch.tensor<[12,77,64],f16>
    %int1_1033 = torch.constant.int 1
    %int12_1034 = torch.constant.int 12
    %int77_1035 = torch.constant.int 77
    %int64_1036 = torch.constant.int 64
    %1170 = torch.prim.ListConstruct %int1_1033, %int12_1034, %int77_1035, %int64_1036 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1171 = torch.aten.view %1169, %1170 : !torch.tensor<[12,77,64],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,64],f16>
    %int1_1037 = torch.constant.int 1
    %int2_1038 = torch.constant.int 2
    %1172 = torch.aten.transpose.int %1171, %int1_1037, %int2_1038 : !torch.tensor<[1,12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int0_1039 = torch.constant.int 0
    %1173 = torch.aten.clone %1172, %int0_1039 : !torch.tensor<[1,77,12,64],f16>, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int1_1040 = torch.constant.int 1
    %int77_1041 = torch.constant.int 77
    %int768_1042 = torch.constant.int 768
    %1174 = torch.prim.ListConstruct %int1_1040, %int77_1041, %int768_1042 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1175 = torch.aten._unsafe_view %1173, %1174 : !torch.tensor<[1,77,12,64],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %1176 = torch.aten.t %241 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_1043 = torch.constant.int 77
    %int768_1044 = torch.constant.int 768
    %1177 = torch.prim.ListConstruct %int77_1043, %int768_1044 : (!torch.int, !torch.int) -> !torch.list<int>
    %1178 = torch.aten.view %1175, %1177 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_1045 = torch.constant.int 1
    %int1_1046 = torch.constant.int 1
    %1179 = torch.aten.addmm %240, %1178, %1176, %int1_1045, %int1_1046 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_1047 = torch.constant.int 1
    %int77_1048 = torch.constant.int 77
    %int768_1049 = torch.constant.int 768
    %1180 = torch.prim.ListConstruct %int1_1047, %int77_1048, %int768_1049 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1181 = torch.aten.view %1179, %1180 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_1050 = torch.constant.int 1
    %1182 = torch.aten.add.Tensor %1121, %1181, %int1_1050 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_1051 = torch.constant.int 768
    %1183 = torch.prim.ListConstruct %int768_1051 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_1052 = torch.constant.float 1.000000e-05
    %result0_1053, %result1_1054, %result2_1055 = torch.aten.native_layer_norm %1182, %1183, %239, %238, %float1.000000e-05_1052 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %1184 = torch.aten.t %237 : !torch.tensor -> !torch.tensor<[768,3072],f16>
    %int77_1056 = torch.constant.int 77
    %int768_1057 = torch.constant.int 768
    %1185 = torch.prim.ListConstruct %int77_1056, %int768_1057 : (!torch.int, !torch.int) -> !torch.list<int>
    %1186 = torch.aten.view %result0_1053, %1185 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_1058 = torch.constant.int 1
    %int1_1059 = torch.constant.int 1
    %1187 = torch.aten.addmm %236, %1186, %1184, %int1_1058, %int1_1059 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,3072],f16>, !torch.int, !torch.int -> !torch.tensor<[77,3072],f16>
    %int1_1060 = torch.constant.int 1
    %int77_1061 = torch.constant.int 77
    %int3072_1062 = torch.constant.int 3072
    %1188 = torch.prim.ListConstruct %int1_1060, %int77_1061, %int3072_1062 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1189 = torch.aten.view %1187, %1188 : !torch.tensor<[77,3072],f16>, !torch.list<int> -> !torch.tensor<[1,77,3072],f16>
    %1190 = torch.tensor.literal(dense<1.702000e+00> : tensor<f64>) : !torch.tensor<[],f64>
    %1191 = torch.aten.mul.Tensor %1189, %1190 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,3072],f16>
    %1192 = torch.aten.sigmoid %1191 : !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %1193 = torch.aten.mul.Tensor %1189, %1192 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %1194 = torch.aten.t %235 : !torch.tensor -> !torch.tensor<[3072,768],f16>
    %int77_1063 = torch.constant.int 77
    %int3072_1064 = torch.constant.int 3072
    %1195 = torch.prim.ListConstruct %int77_1063, %int3072_1064 : (!torch.int, !torch.int) -> !torch.list<int>
    %1196 = torch.aten.view %1193, %1195 : !torch.tensor<[1,77,3072],f16>, !torch.list<int> -> !torch.tensor<[77,3072],f16>
    %int1_1065 = torch.constant.int 1
    %int1_1066 = torch.constant.int 1
    %1197 = torch.aten.addmm %234, %1196, %1194, %int1_1065, %int1_1066 : !torch.tensor, !torch.tensor<[77,3072],f16>, !torch.tensor<[3072,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_1067 = torch.constant.int 1
    %int77_1068 = torch.constant.int 77
    %int768_1069 = torch.constant.int 768
    %1198 = torch.prim.ListConstruct %int1_1067, %int77_1068, %int768_1069 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1199 = torch.aten.view %1197, %1198 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_1070 = torch.constant.int 1
    %1200 = torch.aten.add.Tensor %1182, %1199, %int1_1070 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_1071 = torch.constant.int 768
    %1201 = torch.prim.ListConstruct %int768_1071 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_1072 = torch.constant.float 1.000000e-05
    %result0_1073, %result1_1074, %result2_1075 = torch.aten.native_layer_norm %1200, %1201, %233, %232, %float1.000000e-05_1072 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %1202 = torch.aten.t %231 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_1076 = torch.constant.int 77
    %int768_1077 = torch.constant.int 768
    %1203 = torch.prim.ListConstruct %int77_1076, %int768_1077 : (!torch.int, !torch.int) -> !torch.list<int>
    %1204 = torch.aten.view %result0_1073, %1203 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_1078 = torch.constant.int 1
    %int1_1079 = torch.constant.int 1
    %1205 = torch.aten.addmm %230, %1204, %1202, %int1_1078, %int1_1079 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_1080 = torch.constant.int 1
    %int77_1081 = torch.constant.int 77
    %int768_1082 = torch.constant.int 768
    %1206 = torch.prim.ListConstruct %int1_1080, %int77_1081, %int768_1082 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1207 = torch.aten.view %1205, %1206 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %1208 = torch.tensor.literal(dense<1.250000e-01> : tensor<f64>) : !torch.tensor<[],f64>
    %1209 = torch.aten.mul.Tensor %1207, %1208 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,768],f16>
    %1210 = torch.aten.t %229 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_1083 = torch.constant.int 77
    %int768_1084 = torch.constant.int 768
    %1211 = torch.prim.ListConstruct %int77_1083, %int768_1084 : (!torch.int, !torch.int) -> !torch.list<int>
    %1212 = torch.aten.view %result0_1073, %1211 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_1085 = torch.constant.int 1
    %int1_1086 = torch.constant.int 1
    %1213 = torch.aten.addmm %228, %1212, %1210, %int1_1085, %int1_1086 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_1087 = torch.constant.int 1
    %int77_1088 = torch.constant.int 77
    %int768_1089 = torch.constant.int 768
    %1214 = torch.prim.ListConstruct %int1_1087, %int77_1088, %int768_1089 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1215 = torch.aten.view %1213, %1214 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_1090 = torch.constant.int 1
    %int-1_1091 = torch.constant.int -1
    %int12_1092 = torch.constant.int 12
    %int64_1093 = torch.constant.int 64
    %1216 = torch.prim.ListConstruct %int1_1090, %int-1_1091, %int12_1092, %int64_1093 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1217 = torch.aten.view %1215, %1216 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_1094 = torch.constant.int 1
    %int2_1095 = torch.constant.int 2
    %1218 = torch.aten.transpose.int %1217, %int1_1094, %int2_1095 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_1096 = torch.constant.int 0
    %1219 = torch.aten.clone %1218, %int0_1096 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %1220 = torch.aten.t %227 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_1097 = torch.constant.int 77
    %int768_1098 = torch.constant.int 768
    %1221 = torch.prim.ListConstruct %int77_1097, %int768_1098 : (!torch.int, !torch.int) -> !torch.list<int>
    %1222 = torch.aten.view %result0_1073, %1221 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_1099 = torch.constant.int 1
    %int1_1100 = torch.constant.int 1
    %1223 = torch.aten.addmm %226, %1222, %1220, %int1_1099, %int1_1100 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_1101 = torch.constant.int 1
    %int77_1102 = torch.constant.int 77
    %int768_1103 = torch.constant.int 768
    %1224 = torch.prim.ListConstruct %int1_1101, %int77_1102, %int768_1103 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1225 = torch.aten.view %1223, %1224 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_1104 = torch.constant.int 1
    %int-1_1105 = torch.constant.int -1
    %int12_1106 = torch.constant.int 12
    %int64_1107 = torch.constant.int 64
    %1226 = torch.prim.ListConstruct %int1_1104, %int-1_1105, %int12_1106, %int64_1107 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1227 = torch.aten.view %1225, %1226 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_1108 = torch.constant.int 1
    %int2_1109 = torch.constant.int 2
    %1228 = torch.aten.transpose.int %1227, %int1_1108, %int2_1109 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_1110 = torch.constant.int 0
    %1229 = torch.aten.clone %1228, %int0_1110 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int1_1111 = torch.constant.int 1
    %int77_1112 = torch.constant.int 77
    %int12_1113 = torch.constant.int 12
    %int64_1114 = torch.constant.int 64
    %1230 = torch.prim.ListConstruct %int1_1111, %int77_1112, %int12_1113, %int64_1114 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1231 = torch.aten.view %1209, %1230 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_1115 = torch.constant.int 1
    %int2_1116 = torch.constant.int 2
    %1232 = torch.aten.transpose.int %1231, %int1_1115, %int2_1116 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_1117 = torch.constant.int 0
    %1233 = torch.aten.clone %1232, %int0_1117 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int12_1118 = torch.constant.int 12
    %int-1_1119 = torch.constant.int -1
    %int64_1120 = torch.constant.int 64
    %1234 = torch.prim.ListConstruct %int12_1118, %int-1_1119, %int64_1120 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1235 = torch.aten.view %1233, %1234 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_1121 = torch.constant.int 12
    %int-1_1122 = torch.constant.int -1
    %int64_1123 = torch.constant.int 64
    %1236 = torch.prim.ListConstruct %int12_1121, %int-1_1122, %int64_1123 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1237 = torch.aten.view %1219, %1236 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_1124 = torch.constant.int 12
    %int-1_1125 = torch.constant.int -1
    %int64_1126 = torch.constant.int 64
    %1238 = torch.prim.ListConstruct %int12_1124, %int-1_1125, %int64_1126 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1239 = torch.aten.view %1229, %1238 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int1_1127 = torch.constant.int 1
    %int2_1128 = torch.constant.int 2
    %1240 = torch.aten.transpose.int %1237, %int1_1127, %int2_1128 : !torch.tensor<[12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[12,64,77],f16>
    %1241 = torch.aten.bmm %1235, %1240 : !torch.tensor<[12,77,64],f16>, !torch.tensor<[12,64,77],f16> -> !torch.tensor<[12,77,77],f16>
    %int1_1129 = torch.constant.int 1
    %int12_1130 = torch.constant.int 12
    %int77_1131 = torch.constant.int 77
    %int77_1132 = torch.constant.int 77
    %1242 = torch.prim.ListConstruct %int1_1129, %int12_1130, %int77_1131, %int77_1132 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1243 = torch.aten.view %1241, %1242 : !torch.tensor<[12,77,77],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,77],f16>
    %int1_1133 = torch.constant.int 1
    %1244 = torch.aten.add.Tensor %1243, %410, %int1_1133 : !torch.tensor<[1,12,77,77],f16>, !torch.tensor<[1,1,77,77],f16>, !torch.int -> !torch.tensor<[1,12,77,77],f16>
    %int12_1134 = torch.constant.int 12
    %int77_1135 = torch.constant.int 77
    %int77_1136 = torch.constant.int 77
    %1245 = torch.prim.ListConstruct %int12_1134, %int77_1135, %int77_1136 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1246 = torch.aten.view %1244, %1245 : !torch.tensor<[1,12,77,77],f16>, !torch.list<int> -> !torch.tensor<[12,77,77],f16>
    %int-1_1137 = torch.constant.int -1
    %false_1138 = torch.constant.bool false
    %1247 = torch.aten._softmax %1246, %int-1_1137, %false_1138 : !torch.tensor<[12,77,77],f16>, !torch.int, !torch.bool -> !torch.tensor<[12,77,77],f16>
    %1248 = torch.aten.bmm %1247, %1239 : !torch.tensor<[12,77,77],f16>, !torch.tensor<[12,77,64],f16> -> !torch.tensor<[12,77,64],f16>
    %int1_1139 = torch.constant.int 1
    %int12_1140 = torch.constant.int 12
    %int77_1141 = torch.constant.int 77
    %int64_1142 = torch.constant.int 64
    %1249 = torch.prim.ListConstruct %int1_1139, %int12_1140, %int77_1141, %int64_1142 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1250 = torch.aten.view %1248, %1249 : !torch.tensor<[12,77,64],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,64],f16>
    %int1_1143 = torch.constant.int 1
    %int2_1144 = torch.constant.int 2
    %1251 = torch.aten.transpose.int %1250, %int1_1143, %int2_1144 : !torch.tensor<[1,12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int0_1145 = torch.constant.int 0
    %1252 = torch.aten.clone %1251, %int0_1145 : !torch.tensor<[1,77,12,64],f16>, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int1_1146 = torch.constant.int 1
    %int77_1147 = torch.constant.int 77
    %int768_1148 = torch.constant.int 768
    %1253 = torch.prim.ListConstruct %int1_1146, %int77_1147, %int768_1148 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1254 = torch.aten._unsafe_view %1252, %1253 : !torch.tensor<[1,77,12,64],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %1255 = torch.aten.t %225 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_1149 = torch.constant.int 77
    %int768_1150 = torch.constant.int 768
    %1256 = torch.prim.ListConstruct %int77_1149, %int768_1150 : (!torch.int, !torch.int) -> !torch.list<int>
    %1257 = torch.aten.view %1254, %1256 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_1151 = torch.constant.int 1
    %int1_1152 = torch.constant.int 1
    %1258 = torch.aten.addmm %224, %1257, %1255, %int1_1151, %int1_1152 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_1153 = torch.constant.int 1
    %int77_1154 = torch.constant.int 77
    %int768_1155 = torch.constant.int 768
    %1259 = torch.prim.ListConstruct %int1_1153, %int77_1154, %int768_1155 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1260 = torch.aten.view %1258, %1259 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_1156 = torch.constant.int 1
    %1261 = torch.aten.add.Tensor %1200, %1260, %int1_1156 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_1157 = torch.constant.int 768
    %1262 = torch.prim.ListConstruct %int768_1157 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_1158 = torch.constant.float 1.000000e-05
    %result0_1159, %result1_1160, %result2_1161 = torch.aten.native_layer_norm %1261, %1262, %223, %222, %float1.000000e-05_1158 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %1263 = torch.aten.t %221 : !torch.tensor -> !torch.tensor<[768,3072],f16>
    %int77_1162 = torch.constant.int 77
    %int768_1163 = torch.constant.int 768
    %1264 = torch.prim.ListConstruct %int77_1162, %int768_1163 : (!torch.int, !torch.int) -> !torch.list<int>
    %1265 = torch.aten.view %result0_1159, %1264 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_1164 = torch.constant.int 1
    %int1_1165 = torch.constant.int 1
    %1266 = torch.aten.addmm %220, %1265, %1263, %int1_1164, %int1_1165 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,3072],f16>, !torch.int, !torch.int -> !torch.tensor<[77,3072],f16>
    %int1_1166 = torch.constant.int 1
    %int77_1167 = torch.constant.int 77
    %int3072_1168 = torch.constant.int 3072
    %1267 = torch.prim.ListConstruct %int1_1166, %int77_1167, %int3072_1168 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1268 = torch.aten.view %1266, %1267 : !torch.tensor<[77,3072],f16>, !torch.list<int> -> !torch.tensor<[1,77,3072],f16>
    %1269 = torch.tensor.literal(dense<1.702000e+00> : tensor<f64>) : !torch.tensor<[],f64>
    %1270 = torch.aten.mul.Tensor %1268, %1269 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,3072],f16>
    %1271 = torch.aten.sigmoid %1270 : !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %1272 = torch.aten.mul.Tensor %1268, %1271 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %1273 = torch.aten.t %219 : !torch.tensor -> !torch.tensor<[3072,768],f16>
    %int77_1169 = torch.constant.int 77
    %int3072_1170 = torch.constant.int 3072
    %1274 = torch.prim.ListConstruct %int77_1169, %int3072_1170 : (!torch.int, !torch.int) -> !torch.list<int>
    %1275 = torch.aten.view %1272, %1274 : !torch.tensor<[1,77,3072],f16>, !torch.list<int> -> !torch.tensor<[77,3072],f16>
    %int1_1171 = torch.constant.int 1
    %int1_1172 = torch.constant.int 1
    %1276 = torch.aten.addmm %218, %1275, %1273, %int1_1171, %int1_1172 : !torch.tensor, !torch.tensor<[77,3072],f16>, !torch.tensor<[3072,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_1173 = torch.constant.int 1
    %int77_1174 = torch.constant.int 77
    %int768_1175 = torch.constant.int 768
    %1277 = torch.prim.ListConstruct %int1_1173, %int77_1174, %int768_1175 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1278 = torch.aten.view %1276, %1277 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_1176 = torch.constant.int 1
    %1279 = torch.aten.add.Tensor %1261, %1278, %int1_1176 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_1177 = torch.constant.int 768
    %1280 = torch.prim.ListConstruct %int768_1177 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_1178 = torch.constant.float 1.000000e-05
    %result0_1179, %result1_1180, %result2_1181 = torch.aten.native_layer_norm %1279, %1280, %217, %216, %float1.000000e-05_1178 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %1281 = torch.aten.t %215 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_1182 = torch.constant.int 77
    %int768_1183 = torch.constant.int 768
    %1282 = torch.prim.ListConstruct %int77_1182, %int768_1183 : (!torch.int, !torch.int) -> !torch.list<int>
    %1283 = torch.aten.view %result0_1179, %1282 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_1184 = torch.constant.int 1
    %int1_1185 = torch.constant.int 1
    %1284 = torch.aten.addmm %214, %1283, %1281, %int1_1184, %int1_1185 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_1186 = torch.constant.int 1
    %int77_1187 = torch.constant.int 77
    %int768_1188 = torch.constant.int 768
    %1285 = torch.prim.ListConstruct %int1_1186, %int77_1187, %int768_1188 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1286 = torch.aten.view %1284, %1285 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %1287 = torch.tensor.literal(dense<1.250000e-01> : tensor<f64>) : !torch.tensor<[],f64>
    %1288 = torch.aten.mul.Tensor %1286, %1287 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,768],f16>
    %1289 = torch.aten.t %213 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_1189 = torch.constant.int 77
    %int768_1190 = torch.constant.int 768
    %1290 = torch.prim.ListConstruct %int77_1189, %int768_1190 : (!torch.int, !torch.int) -> !torch.list<int>
    %1291 = torch.aten.view %result0_1179, %1290 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_1191 = torch.constant.int 1
    %int1_1192 = torch.constant.int 1
    %1292 = torch.aten.addmm %212, %1291, %1289, %int1_1191, %int1_1192 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_1193 = torch.constant.int 1
    %int77_1194 = torch.constant.int 77
    %int768_1195 = torch.constant.int 768
    %1293 = torch.prim.ListConstruct %int1_1193, %int77_1194, %int768_1195 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1294 = torch.aten.view %1292, %1293 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_1196 = torch.constant.int 1
    %int-1_1197 = torch.constant.int -1
    %int12_1198 = torch.constant.int 12
    %int64_1199 = torch.constant.int 64
    %1295 = torch.prim.ListConstruct %int1_1196, %int-1_1197, %int12_1198, %int64_1199 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1296 = torch.aten.view %1294, %1295 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_1200 = torch.constant.int 1
    %int2_1201 = torch.constant.int 2
    %1297 = torch.aten.transpose.int %1296, %int1_1200, %int2_1201 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_1202 = torch.constant.int 0
    %1298 = torch.aten.clone %1297, %int0_1202 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %1299 = torch.aten.t %211 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_1203 = torch.constant.int 77
    %int768_1204 = torch.constant.int 768
    %1300 = torch.prim.ListConstruct %int77_1203, %int768_1204 : (!torch.int, !torch.int) -> !torch.list<int>
    %1301 = torch.aten.view %result0_1179, %1300 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_1205 = torch.constant.int 1
    %int1_1206 = torch.constant.int 1
    %1302 = torch.aten.addmm %210, %1301, %1299, %int1_1205, %int1_1206 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_1207 = torch.constant.int 1
    %int77_1208 = torch.constant.int 77
    %int768_1209 = torch.constant.int 768
    %1303 = torch.prim.ListConstruct %int1_1207, %int77_1208, %int768_1209 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1304 = torch.aten.view %1302, %1303 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_1210 = torch.constant.int 1
    %int-1_1211 = torch.constant.int -1
    %int12_1212 = torch.constant.int 12
    %int64_1213 = torch.constant.int 64
    %1305 = torch.prim.ListConstruct %int1_1210, %int-1_1211, %int12_1212, %int64_1213 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1306 = torch.aten.view %1304, %1305 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_1214 = torch.constant.int 1
    %int2_1215 = torch.constant.int 2
    %1307 = torch.aten.transpose.int %1306, %int1_1214, %int2_1215 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_1216 = torch.constant.int 0
    %1308 = torch.aten.clone %1307, %int0_1216 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int1_1217 = torch.constant.int 1
    %int77_1218 = torch.constant.int 77
    %int12_1219 = torch.constant.int 12
    %int64_1220 = torch.constant.int 64
    %1309 = torch.prim.ListConstruct %int1_1217, %int77_1218, %int12_1219, %int64_1220 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1310 = torch.aten.view %1288, %1309 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,12,64],f16>
    %int1_1221 = torch.constant.int 1
    %int2_1222 = torch.constant.int 2
    %1311 = torch.aten.transpose.int %1310, %int1_1221, %int2_1222 : !torch.tensor<[1,77,12,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int0_1223 = torch.constant.int 0
    %1312 = torch.aten.clone %1311, %int0_1223 : !torch.tensor<[1,12,77,64],f16>, !torch.int -> !torch.tensor<[1,12,77,64],f16>
    %int12_1224 = torch.constant.int 12
    %int-1_1225 = torch.constant.int -1
    %int64_1226 = torch.constant.int 64
    %1313 = torch.prim.ListConstruct %int12_1224, %int-1_1225, %int64_1226 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1314 = torch.aten.view %1312, %1313 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_1227 = torch.constant.int 12
    %int-1_1228 = torch.constant.int -1
    %int64_1229 = torch.constant.int 64
    %1315 = torch.prim.ListConstruct %int12_1227, %int-1_1228, %int64_1229 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1316 = torch.aten.view %1298, %1315 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int12_1230 = torch.constant.int 12
    %int-1_1231 = torch.constant.int -1
    %int64_1232 = torch.constant.int 64
    %1317 = torch.prim.ListConstruct %int12_1230, %int-1_1231, %int64_1232 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1318 = torch.aten.view %1308, %1317 : !torch.tensor<[1,12,77,64],f16>, !torch.list<int> -> !torch.tensor<[12,77,64],f16>
    %int1_1233 = torch.constant.int 1
    %int2_1234 = torch.constant.int 2
    %1319 = torch.aten.transpose.int %1316, %int1_1233, %int2_1234 : !torch.tensor<[12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[12,64,77],f16>
    %1320 = torch.aten.bmm %1314, %1319 : !torch.tensor<[12,77,64],f16>, !torch.tensor<[12,64,77],f16> -> !torch.tensor<[12,77,77],f16>
    %int1_1235 = torch.constant.int 1
    %int12_1236 = torch.constant.int 12
    %int77_1237 = torch.constant.int 77
    %int77_1238 = torch.constant.int 77
    %1321 = torch.prim.ListConstruct %int1_1235, %int12_1236, %int77_1237, %int77_1238 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1322 = torch.aten.view %1320, %1321 : !torch.tensor<[12,77,77],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,77],f16>
    %int1_1239 = torch.constant.int 1
    %1323 = torch.aten.add.Tensor %1322, %410, %int1_1239 : !torch.tensor<[1,12,77,77],f16>, !torch.tensor<[1,1,77,77],f16>, !torch.int -> !torch.tensor<[1,12,77,77],f16>
    %int12_1240 = torch.constant.int 12
    %int77_1241 = torch.constant.int 77
    %int77_1242 = torch.constant.int 77
    %1324 = torch.prim.ListConstruct %int12_1240, %int77_1241, %int77_1242 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1325 = torch.aten.view %1323, %1324 : !torch.tensor<[1,12,77,77],f16>, !torch.list<int> -> !torch.tensor<[12,77,77],f16>
    %int-1_1243 = torch.constant.int -1
    %false_1244 = torch.constant.bool false
    %1326 = torch.aten._softmax %1325, %int-1_1243, %false_1244 : !torch.tensor<[12,77,77],f16>, !torch.int, !torch.bool -> !torch.tensor<[12,77,77],f16>
    %1327 = torch.aten.bmm %1326, %1318 : !torch.tensor<[12,77,77],f16>, !torch.tensor<[12,77,64],f16> -> !torch.tensor<[12,77,64],f16>
    %int1_1245 = torch.constant.int 1
    %int12_1246 = torch.constant.int 12
    %int77_1247 = torch.constant.int 77
    %int64_1248 = torch.constant.int 64
    %1328 = torch.prim.ListConstruct %int1_1245, %int12_1246, %int77_1247, %int64_1248 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1329 = torch.aten.view %1327, %1328 : !torch.tensor<[12,77,64],f16>, !torch.list<int> -> !torch.tensor<[1,12,77,64],f16>
    %int1_1249 = torch.constant.int 1
    %int2_1250 = torch.constant.int 2
    %1330 = torch.aten.transpose.int %1329, %int1_1249, %int2_1250 : !torch.tensor<[1,12,77,64],f16>, !torch.int, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int0_1251 = torch.constant.int 0
    %1331 = torch.aten.clone %1330, %int0_1251 : !torch.tensor<[1,77,12,64],f16>, !torch.int -> !torch.tensor<[1,77,12,64],f16>
    %int1_1252 = torch.constant.int 1
    %int77_1253 = torch.constant.int 77
    %int768_1254 = torch.constant.int 768
    %1332 = torch.prim.ListConstruct %int1_1252, %int77_1253, %int768_1254 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1333 = torch.aten._unsafe_view %1331, %1332 : !torch.tensor<[1,77,12,64],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %1334 = torch.aten.t %209 : !torch.tensor -> !torch.tensor<[768,768],f16>
    %int77_1255 = torch.constant.int 77
    %int768_1256 = torch.constant.int 768
    %1335 = torch.prim.ListConstruct %int77_1255, %int768_1256 : (!torch.int, !torch.int) -> !torch.list<int>
    %1336 = torch.aten.view %1333, %1335 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_1257 = torch.constant.int 1
    %int1_1258 = torch.constant.int 1
    %1337 = torch.aten.addmm %208, %1336, %1334, %int1_1257, %int1_1258 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_1259 = torch.constant.int 1
    %int77_1260 = torch.constant.int 77
    %int768_1261 = torch.constant.int 768
    %1338 = torch.prim.ListConstruct %int1_1259, %int77_1260, %int768_1261 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1339 = torch.aten.view %1337, %1338 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_1262 = torch.constant.int 1
    %1340 = torch.aten.add.Tensor %1279, %1339, %int1_1262 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_1263 = torch.constant.int 768
    %1341 = torch.prim.ListConstruct %int768_1263 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_1264 = torch.constant.float 1.000000e-05
    %result0_1265, %result1_1266, %result2_1267 = torch.aten.native_layer_norm %1340, %1341, %207, %206, %float1.000000e-05_1264 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %1342 = torch.aten.t %205 : !torch.tensor -> !torch.tensor<[768,3072],f16>
    %int77_1268 = torch.constant.int 77
    %int768_1269 = torch.constant.int 768
    %1343 = torch.prim.ListConstruct %int77_1268, %int768_1269 : (!torch.int, !torch.int) -> !torch.list<int>
    %1344 = torch.aten.view %result0_1265, %1343 : !torch.tensor<[1,77,768],f16>, !torch.list<int> -> !torch.tensor<[77,768],f16>
    %int1_1270 = torch.constant.int 1
    %int1_1271 = torch.constant.int 1
    %1345 = torch.aten.addmm %204, %1344, %1342, %int1_1270, %int1_1271 : !torch.tensor, !torch.tensor<[77,768],f16>, !torch.tensor<[768,3072],f16>, !torch.int, !torch.int -> !torch.tensor<[77,3072],f16>
    %int1_1272 = torch.constant.int 1
    %int77_1273 = torch.constant.int 77
    %int3072_1274 = torch.constant.int 3072
    %1346 = torch.prim.ListConstruct %int1_1272, %int77_1273, %int3072_1274 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1347 = torch.aten.view %1345, %1346 : !torch.tensor<[77,3072],f16>, !torch.list<int> -> !torch.tensor<[1,77,3072],f16>
    %1348 = torch.tensor.literal(dense<1.702000e+00> : tensor<f64>) : !torch.tensor<[],f64>
    %1349 = torch.aten.mul.Tensor %1347, %1348 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[],f64> -> !torch.tensor<[1,77,3072],f16>
    %1350 = torch.aten.sigmoid %1349 : !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %1351 = torch.aten.mul.Tensor %1347, %1350 : !torch.tensor<[1,77,3072],f16>, !torch.tensor<[1,77,3072],f16> -> !torch.tensor<[1,77,3072],f16>
    %1352 = torch.aten.t %203 : !torch.tensor -> !torch.tensor<[3072,768],f16>
    %int77_1275 = torch.constant.int 77
    %int3072_1276 = torch.constant.int 3072
    %1353 = torch.prim.ListConstruct %int77_1275, %int3072_1276 : (!torch.int, !torch.int) -> !torch.list<int>
    %1354 = torch.aten.view %1351, %1353 : !torch.tensor<[1,77,3072],f16>, !torch.list<int> -> !torch.tensor<[77,3072],f16>
    %int1_1277 = torch.constant.int 1
    %int1_1278 = torch.constant.int 1
    %1355 = torch.aten.addmm %202, %1354, %1352, %int1_1277, %int1_1278 : !torch.tensor, !torch.tensor<[77,3072],f16>, !torch.tensor<[3072,768],f16>, !torch.int, !torch.int -> !torch.tensor<[77,768],f16>
    %int1_1279 = torch.constant.int 1
    %int77_1280 = torch.constant.int 77
    %int768_1281 = torch.constant.int 768
    %1356 = torch.prim.ListConstruct %int1_1279, %int77_1280, %int768_1281 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int>
    %1357 = torch.aten.view %1355, %1356 : !torch.tensor<[77,768],f16>, !torch.list<int> -> !torch.tensor<[1,77,768],f16>
    %int1_1282 = torch.constant.int 1
    %1358 = torch.aten.add.Tensor %1340, %1357, %int1_1282 : !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,768],f16>, !torch.int -> !torch.tensor<[1,77,768],f16>
    %int768_1283 = torch.constant.int 768
    %1359 = torch.prim.ListConstruct %int768_1283 : (!torch.int) -> !torch.list<int>
    %float1.000000e-05_1284 = torch.constant.float 1.000000e-05
    %result0_1285, %result1_1286, %result2_1287 = torch.aten.native_layer_norm %1358, %1359, %201, %200, %float1.000000e-05_1284 : !torch.tensor<[1,77,768],f16>, !torch.list<int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor<[1,77,768],f16>, !torch.tensor<[1,77,1],f32>, !torch.tensor<[1,77,1],f32>
    %1360 = torch.tensor_static_info_cast %result0_1285 : !torch.tensor<[1,77,768],f16> to !torch.tensor
    return %1360 : !torch.tensor
  }
  torch.class_type @__torch__.torch.fx.graph_module._lambda {
    torch.attr private "_param_constant0" : !torch.tensor
    torch.attr private "_param_constant1" : !torch.tensor
    torch.attr private "_param_constant2" : !torch.tensor
    torch.attr private "_param_constant3" : !torch.tensor
    torch.attr private "_param_constant4" : !torch.tensor
    torch.attr private "_param_constant5" : !torch.tensor
    torch.attr private "_param_constant6" : !torch.tensor
    torch.attr private "_param_constant7" : !torch.tensor
    torch.attr private "_param_constant8" : !torch.tensor
    torch.attr private "_param_constant9" : !torch.tensor
    torch.attr private "_param_constant10" : !torch.tensor
    torch.attr private "_param_constant11" : !torch.tensor
    torch.attr private "_param_constant12" : !torch.tensor
    torch.attr private "_param_constant13" : !torch.tensor
    torch.attr private "_param_constant14" : !torch.tensor
    torch.attr private "_param_constant15" : !torch.tensor
    torch.attr private "_param_constant16" : !torch.tensor
    torch.attr private "_param_constant17" : !torch.tensor
    torch.attr private "_param_constant18" : !torch.tensor
    torch.attr private "_param_constant19" : !torch.tensor
    torch.attr private "_param_constant20" : !torch.tensor
    torch.attr private "_param_constant21" : !torch.tensor
    torch.attr private "_param_constant22" : !torch.tensor
    torch.attr private "_param_constant23" : !torch.tensor
    torch.attr private "_param_constant24" : !torch.tensor
    torch.attr private "_param_constant25" : !torch.tensor
    torch.attr private "_param_constant26" : !torch.tensor
    torch.attr private "_param_constant27" : !torch.tensor
    torch.attr private "_param_constant28" : !torch.tensor
    torch.attr private "_param_constant29" : !torch.tensor
    torch.attr private "_param_constant30" : !torch.tensor
    torch.attr private "_param_constant31" : !torch.tensor
    torch.attr private "_param_constant32" : !torch.tensor
    torch.attr private "_param_constant33" : !torch.tensor
    torch.attr private "_param_constant34" : !torch.tensor
    torch.attr private "_param_constant35" : !torch.tensor
    torch.attr private "_param_constant36" : !torch.tensor
    torch.attr private "_param_constant37" : !torch.tensor
    torch.attr private "_param_constant38" : !torch.tensor
    torch.attr private "_param_constant39" : !torch.tensor
    torch.attr private "_param_constant40" : !torch.tensor
    torch.attr private "_param_constant41" : !torch.tensor
    torch.attr private "_param_constant42" : !torch.tensor
    torch.attr private "_param_constant43" : !torch.tensor
    torch.attr private "_param_constant44" : !torch.tensor
    torch.attr private "_param_constant45" : !torch.tensor
    torch.attr private "_param_constant46" : !torch.tensor
    torch.attr private "_param_constant47" : !torch.tensor
    torch.attr private "_param_constant48" : !torch.tensor
    torch.attr private "_param_constant49" : !torch.tensor
    torch.attr private "_param_constant50" : !torch.tensor
    torch.attr private "_param_constant51" : !torch.tensor
    torch.attr private "_param_constant52" : !torch.tensor
    torch.attr private "_param_constant53" : !torch.tensor
    torch.attr private "_param_constant54" : !torch.tensor
    torch.attr private "_param_constant55" : !torch.tensor
    torch.attr private "_param_constant56" : !torch.tensor
    torch.attr private "_param_constant57" : !torch.tensor
    torch.attr private "_param_constant58" : !torch.tensor
    torch.attr private "_param_constant59" : !torch.tensor
    torch.attr private "_param_constant60" : !torch.tensor
    torch.attr private "_param_constant61" : !torch.tensor
    torch.attr private "_param_constant62" : !torch.tensor
    torch.attr private "_param_constant63" : !torch.tensor
    torch.attr private "_param_constant64" : !torch.tensor
    torch.attr private "_param_constant65" : !torch.tensor
    torch.attr private "_param_constant66" : !torch.tensor
    torch.attr private "_param_constant67" : !torch.tensor
    torch.attr private "_param_constant68" : !torch.tensor
    torch.attr private "_param_constant69" : !torch.tensor
    torch.attr private "_param_constant70" : !torch.tensor
    torch.attr private "_param_constant71" : !torch.tensor
    torch.attr private "_param_constant72" : !torch.tensor
    torch.attr private "_param_constant73" : !torch.tensor
    torch.attr private "_param_constant74" : !torch.tensor
    torch.attr private "_param_constant75" : !torch.tensor
    torch.attr private "_param_constant76" : !torch.tensor
    torch.attr private "_param_constant77" : !torch.tensor
    torch.attr private "_param_constant78" : !torch.tensor
    torch.attr private "_param_constant79" : !torch.tensor
    torch.attr private "_param_constant80" : !torch.tensor
    torch.attr private "_param_constant81" : !torch.tensor
    torch.attr private "_param_constant82" : !torch.tensor
    torch.attr private "_param_constant83" : !torch.tensor
    torch.attr private "_param_constant84" : !torch.tensor
    torch.attr private "_param_constant85" : !torch.tensor
    torch.attr private "_param_constant86" : !torch.tensor
    torch.attr private "_param_constant87" : !torch.tensor
    torch.attr private "_param_constant88" : !torch.tensor
    torch.attr private "_param_constant89" : !torch.tensor
    torch.attr private "_param_constant90" : !torch.tensor
    torch.attr private "_param_constant91" : !torch.tensor
    torch.attr private "_param_constant92" : !torch.tensor
    torch.attr private "_param_constant93" : !torch.tensor
    torch.attr private "_param_constant94" : !torch.tensor
    torch.attr private "_param_constant95" : !torch.tensor
    torch.attr private "_param_constant96" : !torch.tensor
    torch.attr private "_param_constant97" : !torch.tensor
    torch.attr private "_param_constant98" : !torch.tensor
    torch.attr private "_param_constant99" : !torch.tensor
    torch.attr private "_param_constant100" : !torch.tensor
    torch.attr private "_param_constant101" : !torch.tensor
    torch.attr private "_param_constant102" : !torch.tensor
    torch.attr private "_param_constant103" : !torch.tensor
    torch.attr private "_param_constant104" : !torch.tensor
    torch.attr private "_param_constant105" : !torch.tensor
    torch.attr private "_param_constant106" : !torch.tensor
    torch.attr private "_param_constant107" : !torch.tensor
    torch.attr private "_param_constant108" : !torch.tensor
    torch.attr private "_param_constant109" : !torch.tensor
    torch.attr private "_param_constant110" : !torch.tensor
    torch.attr private "_param_constant111" : !torch.tensor
    torch.attr private "_param_constant112" : !torch.tensor
    torch.attr private "_param_constant113" : !torch.tensor
    torch.attr private "_param_constant114" : !torch.tensor
    torch.attr private "_param_constant115" : !torch.tensor
    torch.attr private "_param_constant116" : !torch.tensor
    torch.attr private "_param_constant117" : !torch.tensor
    torch.attr private "_param_constant118" : !torch.tensor
    torch.attr private "_param_constant119" : !torch.tensor
    torch.attr private "_param_constant120" : !torch.tensor
    torch.attr private "_param_constant121" : !torch.tensor
    torch.attr private "_param_constant122" : !torch.tensor
    torch.attr private "_param_constant123" : !torch.tensor
    torch.attr private "_param_constant124" : !torch.tensor
    torch.attr private "_param_constant125" : !torch.tensor
    torch.attr private "_param_constant126" : !torch.tensor
    torch.attr private "_param_constant127" : !torch.tensor
    torch.attr private "_param_constant128" : !torch.tensor
    torch.attr private "_param_constant129" : !torch.tensor
    torch.attr private "_param_constant130" : !torch.tensor
    torch.attr private "_param_constant131" : !torch.tensor
    torch.attr private "_param_constant132" : !torch.tensor
    torch.attr private "_param_constant133" : !torch.tensor
    torch.attr private "_param_constant134" : !torch.tensor
    torch.attr private "_param_constant135" : !torch.tensor
    torch.attr private "_param_constant136" : !torch.tensor
    torch.attr private "_param_constant137" : !torch.tensor
    torch.attr private "_param_constant138" : !torch.tensor
    torch.attr private "_param_constant139" : !torch.tensor
    torch.attr private "_param_constant140" : !torch.tensor
    torch.attr private "_param_constant141" : !torch.tensor
    torch.attr private "_param_constant142" : !torch.tensor
    torch.attr private "_param_constant143" : !torch.tensor
    torch.attr private "_param_constant144" : !torch.tensor
    torch.attr private "_param_constant145" : !torch.tensor
    torch.attr private "_param_constant146" : !torch.tensor
    torch.attr private "_param_constant147" : !torch.tensor
    torch.attr private "_param_constant148" : !torch.tensor
    torch.attr private "_param_constant149" : !torch.tensor
    torch.attr private "_param_constant150" : !torch.tensor
    torch.attr private "_param_constant151" : !torch.tensor
    torch.attr private "_param_constant152" : !torch.tensor
    torch.attr private "_param_constant153" : !torch.tensor
    torch.attr private "_param_constant154" : !torch.tensor
    torch.attr private "_param_constant155" : !torch.tensor
    torch.attr private "_param_constant156" : !torch.tensor
    torch.attr private "_param_constant157" : !torch.tensor
    torch.attr private "_param_constant158" : !torch.tensor
    torch.attr private "_param_constant159" : !torch.tensor
    torch.attr private "_param_constant160" : !torch.tensor
    torch.attr private "_param_constant161" : !torch.tensor
    torch.attr private "_param_constant162" : !torch.tensor
    torch.attr private "_param_constant163" : !torch.tensor
    torch.attr private "_param_constant164" : !torch.tensor
    torch.attr private "_param_constant165" : !torch.tensor
    torch.attr private "_param_constant166" : !torch.tensor
    torch.attr private "_param_constant167" : !torch.tensor
    torch.attr private "_param_constant168" : !torch.tensor
    torch.attr private "_param_constant169" : !torch.tensor
    torch.attr private "_param_constant170" : !torch.tensor
    torch.attr private "_param_constant171" : !torch.tensor
    torch.attr private "_param_constant172" : !torch.tensor
    torch.attr private "_param_constant173" : !torch.tensor
    torch.attr private "_param_constant174" : !torch.tensor
    torch.attr private "_param_constant175" : !torch.tensor
    torch.attr private "_param_constant176" : !torch.tensor
    torch.attr private "_param_constant177" : !torch.tensor
    torch.attr private "_param_constant178" : !torch.tensor
    torch.attr private "_param_constant179" : !torch.tensor
    torch.attr private "_param_constant180" : !torch.tensor
    torch.attr private "_param_constant181" : !torch.tensor
    torch.attr private "_param_constant182" : !torch.tensor
    torch.attr private "_param_constant183" : !torch.tensor
    torch.attr private "_param_constant184" : !torch.tensor
    torch.attr private "_param_constant185" : !torch.tensor
    torch.attr private "_param_constant186" : !torch.tensor
    torch.attr private "_param_constant187" : !torch.tensor
    torch.attr private "_param_constant188" : !torch.tensor
    torch.attr private "_param_constant189" : !torch.tensor
    torch.attr private "_param_constant190" : !torch.tensor
    torch.attr private "_param_constant191" : !torch.tensor
    torch.attr private "_param_constant192" : !torch.tensor
    torch.attr private "_param_constant193" : !torch.tensor
    torch.attr private "_param_constant194" : !torch.tensor
    torch.attr private "_param_constant195" : !torch.tensor
    torch.attr private "_tensor_constant0" : !torch.tensor
    torch.attr private "_tensor_constant1" : !torch.tensor
    torch.attr private "training" : !torch.bool
    torch.attr private "_is_full_backward_hook" : !torch.optional<bool>
    torch.method "forward", @__torch__.torch.fx.graph_module._lambda.forward
  }
  %0 = torch.tensor.literal(dense_resource<__elided__> : tensor<49408x768xf16>) : !torch.tensor<[49408,768],f16>
  %1 = torch.tensor.literal(dense_resource<__elided__> : tensor<77x768xf16>) : !torch.tensor<[77,768],f16>
  %2 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %3 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %4 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %5 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %6 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %7 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %8 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %9 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %10 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %11 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %12 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %13 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %14 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072x768xf16>) : !torch.tensor<[3072,768],f16>
  %15 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072xf16>) : !torch.tensor<[3072],f16>
  %16 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x3072xf16>) : !torch.tensor<[768,3072],f16>
  %17 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %18 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %19 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %20 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %21 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %22 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %23 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %24 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %25 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %26 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %27 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %28 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %29 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %30 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072x768xf16>) : !torch.tensor<[3072,768],f16>
  %31 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072xf16>) : !torch.tensor<[3072],f16>
  %32 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x3072xf16>) : !torch.tensor<[768,3072],f16>
  %33 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %34 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %35 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %36 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %37 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %38 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %39 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %40 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %41 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %42 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %43 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %44 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %45 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %46 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072x768xf16>) : !torch.tensor<[3072,768],f16>
  %47 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072xf16>) : !torch.tensor<[3072],f16>
  %48 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x3072xf16>) : !torch.tensor<[768,3072],f16>
  %49 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %50 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %51 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %52 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %53 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %54 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %55 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %56 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %57 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %58 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %59 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %60 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %61 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %62 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072x768xf16>) : !torch.tensor<[3072,768],f16>
  %63 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072xf16>) : !torch.tensor<[3072],f16>
  %64 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x3072xf16>) : !torch.tensor<[768,3072],f16>
  %65 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %66 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %67 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %68 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %69 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %70 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %71 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %72 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %73 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %74 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %75 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %76 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %77 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %78 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072x768xf16>) : !torch.tensor<[3072,768],f16>
  %79 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072xf16>) : !torch.tensor<[3072],f16>
  %80 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x3072xf16>) : !torch.tensor<[768,3072],f16>
  %81 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %82 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %83 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %84 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %85 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %86 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %87 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %88 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %89 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %90 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %91 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %92 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %93 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %94 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072x768xf16>) : !torch.tensor<[3072,768],f16>
  %95 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072xf16>) : !torch.tensor<[3072],f16>
  %96 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x3072xf16>) : !torch.tensor<[768,3072],f16>
  %97 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %98 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %99 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %100 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %101 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %102 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %103 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %104 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %105 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %106 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %107 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %108 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %109 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %110 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072x768xf16>) : !torch.tensor<[3072,768],f16>
  %111 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072xf16>) : !torch.tensor<[3072],f16>
  %112 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x3072xf16>) : !torch.tensor<[768,3072],f16>
  %113 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %114 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %115 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %116 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %117 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %118 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %119 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %120 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %121 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %122 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %123 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %124 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %125 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %126 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072x768xf16>) : !torch.tensor<[3072,768],f16>
  %127 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072xf16>) : !torch.tensor<[3072],f16>
  %128 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x3072xf16>) : !torch.tensor<[768,3072],f16>
  %129 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %130 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %131 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %132 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %133 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %134 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %135 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %136 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %137 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %138 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %139 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %140 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %141 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %142 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072x768xf16>) : !torch.tensor<[3072,768],f16>
  %143 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072xf16>) : !torch.tensor<[3072],f16>
  %144 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x3072xf16>) : !torch.tensor<[768,3072],f16>
  %145 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %146 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %147 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %148 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %149 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %150 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %151 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %152 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %153 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %154 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %155 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %156 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %157 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %158 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072x768xf16>) : !torch.tensor<[3072,768],f16>
  %159 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072xf16>) : !torch.tensor<[3072],f16>
  %160 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x3072xf16>) : !torch.tensor<[768,3072],f16>
  %161 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %162 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %163 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %164 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %165 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %166 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %167 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %168 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %169 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %170 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %171 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %172 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %173 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %174 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072x768xf16>) : !torch.tensor<[3072,768],f16>
  %175 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072xf16>) : !torch.tensor<[3072],f16>
  %176 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x3072xf16>) : !torch.tensor<[768,3072],f16>
  %177 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %178 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %179 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %180 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %181 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %182 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %183 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %184 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %185 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %186 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x768xf16>) : !torch.tensor<[768,768],f16>
  %187 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %188 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %189 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %190 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072x768xf16>) : !torch.tensor<[3072,768],f16>
  %191 = torch.tensor.literal(dense_resource<__elided__> : tensor<3072xf16>) : !torch.tensor<[3072],f16>
  %192 = torch.tensor.literal(dense_resource<__elided__> : tensor<768x3072xf16>) : !torch.tensor<[768,3072],f16>
  %193 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %194 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %195 = torch.tensor.literal(dense_resource<__elided__> : tensor<768xf16>) : !torch.tensor<[768],f16>
  %196 = torch.tensor.literal(dense_resource<__elided__> : tensor<1x77xsi64>) : !torch.tensor<[1,77],si64>
  %197 = torch.tensor.literal(dense<-6.550400e+04> : tensor<f32>) : !torch.tensor<[],f32>
  %true = torch.constant.bool true
  %none = torch.constant.none
  %198 = torch.nn_module {
    torch.slot "_param_constant0", %0 : !torch.tensor<[49408,768],f16>
    torch.slot "_param_constant1", %1 : !torch.tensor<[77,768],f16>
    torch.slot "_param_constant2", %2 : !torch.tensor<[768],f16>
    torch.slot "_param_constant3", %3 : !torch.tensor<[768],f16>
    torch.slot "_param_constant4", %4 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant5", %5 : !torch.tensor<[768],f16>
    torch.slot "_param_constant6", %6 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant7", %7 : !torch.tensor<[768],f16>
    torch.slot "_param_constant8", %8 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant9", %9 : !torch.tensor<[768],f16>
    torch.slot "_param_constant10", %10 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant11", %11 : !torch.tensor<[768],f16>
    torch.slot "_param_constant12", %12 : !torch.tensor<[768],f16>
    torch.slot "_param_constant13", %13 : !torch.tensor<[768],f16>
    torch.slot "_param_constant14", %14 : !torch.tensor<[3072,768],f16>
    torch.slot "_param_constant15", %15 : !torch.tensor<[3072],f16>
    torch.slot "_param_constant16", %16 : !torch.tensor<[768,3072],f16>
    torch.slot "_param_constant17", %17 : !torch.tensor<[768],f16>
    torch.slot "_param_constant18", %18 : !torch.tensor<[768],f16>
    torch.slot "_param_constant19", %19 : !torch.tensor<[768],f16>
    torch.slot "_param_constant20", %20 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant21", %21 : !torch.tensor<[768],f16>
    torch.slot "_param_constant22", %22 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant23", %23 : !torch.tensor<[768],f16>
    torch.slot "_param_constant24", %24 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant25", %25 : !torch.tensor<[768],f16>
    torch.slot "_param_constant26", %26 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant27", %27 : !torch.tensor<[768],f16>
    torch.slot "_param_constant28", %28 : !torch.tensor<[768],f16>
    torch.slot "_param_constant29", %29 : !torch.tensor<[768],f16>
    torch.slot "_param_constant30", %30 : !torch.tensor<[3072,768],f16>
    torch.slot "_param_constant31", %31 : !torch.tensor<[3072],f16>
    torch.slot "_param_constant32", %32 : !torch.tensor<[768,3072],f16>
    torch.slot "_param_constant33", %33 : !torch.tensor<[768],f16>
    torch.slot "_param_constant34", %34 : !torch.tensor<[768],f16>
    torch.slot "_param_constant35", %35 : !torch.tensor<[768],f16>
    torch.slot "_param_constant36", %36 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant37", %37 : !torch.tensor<[768],f16>
    torch.slot "_param_constant38", %38 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant39", %39 : !torch.tensor<[768],f16>
    torch.slot "_param_constant40", %40 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant41", %41 : !torch.tensor<[768],f16>
    torch.slot "_param_constant42", %42 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant43", %43 : !torch.tensor<[768],f16>
    torch.slot "_param_constant44", %44 : !torch.tensor<[768],f16>
    torch.slot "_param_constant45", %45 : !torch.tensor<[768],f16>
    torch.slot "_param_constant46", %46 : !torch.tensor<[3072,768],f16>
    torch.slot "_param_constant47", %47 : !torch.tensor<[3072],f16>
    torch.slot "_param_constant48", %48 : !torch.tensor<[768,3072],f16>
    torch.slot "_param_constant49", %49 : !torch.tensor<[768],f16>
    torch.slot "_param_constant50", %50 : !torch.tensor<[768],f16>
    torch.slot "_param_constant51", %51 : !torch.tensor<[768],f16>
    torch.slot "_param_constant52", %52 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant53", %53 : !torch.tensor<[768],f16>
    torch.slot "_param_constant54", %54 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant55", %55 : !torch.tensor<[768],f16>
    torch.slot "_param_constant56", %56 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant57", %57 : !torch.tensor<[768],f16>
    torch.slot "_param_constant58", %58 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant59", %59 : !torch.tensor<[768],f16>
    torch.slot "_param_constant60", %60 : !torch.tensor<[768],f16>
    torch.slot "_param_constant61", %61 : !torch.tensor<[768],f16>
    torch.slot "_param_constant62", %62 : !torch.tensor<[3072,768],f16>
    torch.slot "_param_constant63", %63 : !torch.tensor<[3072],f16>
    torch.slot "_param_constant64", %64 : !torch.tensor<[768,3072],f16>
    torch.slot "_param_constant65", %65 : !torch.tensor<[768],f16>
    torch.slot "_param_constant66", %66 : !torch.tensor<[768],f16>
    torch.slot "_param_constant67", %67 : !torch.tensor<[768],f16>
    torch.slot "_param_constant68", %68 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant69", %69 : !torch.tensor<[768],f16>
    torch.slot "_param_constant70", %70 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant71", %71 : !torch.tensor<[768],f16>
    torch.slot "_param_constant72", %72 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant73", %73 : !torch.tensor<[768],f16>
    torch.slot "_param_constant74", %74 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant75", %75 : !torch.tensor<[768],f16>
    torch.slot "_param_constant76", %76 : !torch.tensor<[768],f16>
    torch.slot "_param_constant77", %77 : !torch.tensor<[768],f16>
    torch.slot "_param_constant78", %78 : !torch.tensor<[3072,768],f16>
    torch.slot "_param_constant79", %79 : !torch.tensor<[3072],f16>
    torch.slot "_param_constant80", %80 : !torch.tensor<[768,3072],f16>
    torch.slot "_param_constant81", %81 : !torch.tensor<[768],f16>
    torch.slot "_param_constant82", %82 : !torch.tensor<[768],f16>
    torch.slot "_param_constant83", %83 : !torch.tensor<[768],f16>
    torch.slot "_param_constant84", %84 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant85", %85 : !torch.tensor<[768],f16>
    torch.slot "_param_constant86", %86 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant87", %87 : !torch.tensor<[768],f16>
    torch.slot "_param_constant88", %88 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant89", %89 : !torch.tensor<[768],f16>
    torch.slot "_param_constant90", %90 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant91", %91 : !torch.tensor<[768],f16>
    torch.slot "_param_constant92", %92 : !torch.tensor<[768],f16>
    torch.slot "_param_constant93", %93 : !torch.tensor<[768],f16>
    torch.slot "_param_constant94", %94 : !torch.tensor<[3072,768],f16>
    torch.slot "_param_constant95", %95 : !torch.tensor<[3072],f16>
    torch.slot "_param_constant96", %96 : !torch.tensor<[768,3072],f16>
    torch.slot "_param_constant97", %97 : !torch.tensor<[768],f16>
    torch.slot "_param_constant98", %98 : !torch.tensor<[768],f16>
    torch.slot "_param_constant99", %99 : !torch.tensor<[768],f16>
    torch.slot "_param_constant100", %100 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant101", %101 : !torch.tensor<[768],f16>
    torch.slot "_param_constant102", %102 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant103", %103 : !torch.tensor<[768],f16>
    torch.slot "_param_constant104", %104 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant105", %105 : !torch.tensor<[768],f16>
    torch.slot "_param_constant106", %106 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant107", %107 : !torch.tensor<[768],f16>
    torch.slot "_param_constant108", %108 : !torch.tensor<[768],f16>
    torch.slot "_param_constant109", %109 : !torch.tensor<[768],f16>
    torch.slot "_param_constant110", %110 : !torch.tensor<[3072,768],f16>
    torch.slot "_param_constant111", %111 : !torch.tensor<[3072],f16>
    torch.slot "_param_constant112", %112 : !torch.tensor<[768,3072],f16>
    torch.slot "_param_constant113", %113 : !torch.tensor<[768],f16>
    torch.slot "_param_constant114", %114 : !torch.tensor<[768],f16>
    torch.slot "_param_constant115", %115 : !torch.tensor<[768],f16>
    torch.slot "_param_constant116", %116 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant117", %117 : !torch.tensor<[768],f16>
    torch.slot "_param_constant118", %118 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant119", %119 : !torch.tensor<[768],f16>
    torch.slot "_param_constant120", %120 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant121", %121 : !torch.tensor<[768],f16>
    torch.slot "_param_constant122", %122 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant123", %123 : !torch.tensor<[768],f16>
    torch.slot "_param_constant124", %124 : !torch.tensor<[768],f16>
    torch.slot "_param_constant125", %125 : !torch.tensor<[768],f16>
    torch.slot "_param_constant126", %126 : !torch.tensor<[3072,768],f16>
    torch.slot "_param_constant127", %127 : !torch.tensor<[3072],f16>
    torch.slot "_param_constant128", %128 : !torch.tensor<[768,3072],f16>
    torch.slot "_param_constant129", %129 : !torch.tensor<[768],f16>
    torch.slot "_param_constant130", %130 : !torch.tensor<[768],f16>
    torch.slot "_param_constant131", %131 : !torch.tensor<[768],f16>
    torch.slot "_param_constant132", %132 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant133", %133 : !torch.tensor<[768],f16>
    torch.slot "_param_constant134", %134 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant135", %135 : !torch.tensor<[768],f16>
    torch.slot "_param_constant136", %136 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant137", %137 : !torch.tensor<[768],f16>
    torch.slot "_param_constant138", %138 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant139", %139 : !torch.tensor<[768],f16>
    torch.slot "_param_constant140", %140 : !torch.tensor<[768],f16>
    torch.slot "_param_constant141", %141 : !torch.tensor<[768],f16>
    torch.slot "_param_constant142", %142 : !torch.tensor<[3072,768],f16>
    torch.slot "_param_constant143", %143 : !torch.tensor<[3072],f16>
    torch.slot "_param_constant144", %144 : !torch.tensor<[768,3072],f16>
    torch.slot "_param_constant145", %145 : !torch.tensor<[768],f16>
    torch.slot "_param_constant146", %146 : !torch.tensor<[768],f16>
    torch.slot "_param_constant147", %147 : !torch.tensor<[768],f16>
    torch.slot "_param_constant148", %148 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant149", %149 : !torch.tensor<[768],f16>
    torch.slot "_param_constant150", %150 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant151", %151 : !torch.tensor<[768],f16>
    torch.slot "_param_constant152", %152 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant153", %153 : !torch.tensor<[768],f16>
    torch.slot "_param_constant154", %154 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant155", %155 : !torch.tensor<[768],f16>
    torch.slot "_param_constant156", %156 : !torch.tensor<[768],f16>
    torch.slot "_param_constant157", %157 : !torch.tensor<[768],f16>
    torch.slot "_param_constant158", %158 : !torch.tensor<[3072,768],f16>
    torch.slot "_param_constant159", %159 : !torch.tensor<[3072],f16>
    torch.slot "_param_constant160", %160 : !torch.tensor<[768,3072],f16>
    torch.slot "_param_constant161", %161 : !torch.tensor<[768],f16>
    torch.slot "_param_constant162", %162 : !torch.tensor<[768],f16>
    torch.slot "_param_constant163", %163 : !torch.tensor<[768],f16>
    torch.slot "_param_constant164", %164 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant165", %165 : !torch.tensor<[768],f16>
    torch.slot "_param_constant166", %166 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant167", %167 : !torch.tensor<[768],f16>
    torch.slot "_param_constant168", %168 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant169", %169 : !torch.tensor<[768],f16>
    torch.slot "_param_constant170", %170 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant171", %171 : !torch.tensor<[768],f16>
    torch.slot "_param_constant172", %172 : !torch.tensor<[768],f16>
    torch.slot "_param_constant173", %173 : !torch.tensor<[768],f16>
    torch.slot "_param_constant174", %174 : !torch.tensor<[3072,768],f16>
    torch.slot "_param_constant175", %175 : !torch.tensor<[3072],f16>
    torch.slot "_param_constant176", %176 : !torch.tensor<[768,3072],f16>
    torch.slot "_param_constant177", %177 : !torch.tensor<[768],f16>
    torch.slot "_param_constant178", %178 : !torch.tensor<[768],f16>
    torch.slot "_param_constant179", %179 : !torch.tensor<[768],f16>
    torch.slot "_param_constant180", %180 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant181", %181 : !torch.tensor<[768],f16>
    torch.slot "_param_constant182", %182 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant183", %183 : !torch.tensor<[768],f16>
    torch.slot "_param_constant184", %184 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant185", %185 : !torch.tensor<[768],f16>
    torch.slot "_param_constant186", %186 : !torch.tensor<[768,768],f16>
    torch.slot "_param_constant187", %187 : !torch.tensor<[768],f16>
    torch.slot "_param_constant188", %188 : !torch.tensor<[768],f16>
    torch.slot "_param_constant189", %189 : !torch.tensor<[768],f16>
    torch.slot "_param_constant190", %190 : !torch.tensor<[3072,768],f16>
    torch.slot "_param_constant191", %191 : !torch.tensor<[3072],f16>
    torch.slot "_param_constant192", %192 : !torch.tensor<[768,3072],f16>
    torch.slot "_param_constant193", %193 : !torch.tensor<[768],f16>
    torch.slot "_param_constant194", %194 : !torch.tensor<[768],f16>
    torch.slot "_param_constant195", %195 : !torch.tensor<[768],f16>
    torch.slot "_tensor_constant0", %196 : !torch.tensor<[1,77],si64>
    torch.slot "_tensor_constant1", %197 : !torch.tensor<[],f32>
    torch.slot "training", %true : !torch.bool
    torch.slot "_is_full_backward_hook", %none : !torch.none
  } : !torch.nn.Module<"__torch__.torch.fx.graph_module._lambda">
}