big_bird.mlir

module attributes {torch.debug_module_name = "GraphModule"} {
  func private @__torch__.torch.fx.graph_module.___torch_mangle_0.GraphModule.forward(%arg0: !torch.nn.Module<"__torch__.torch.fx.graph_module.___torch_mangle_0.GraphModule">, %arg1: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg2: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg3: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg4: !torch.tensor {torch.type_bound = !torch.vtensor<[2,1024],f32>}, %arg5: !torch.tensor {torch.type_bound = !torch.vtensor<[50358,1024],f32>}, %arg6: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg7: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg8: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg9: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg10: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg11: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg12: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg13: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg14: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg15: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg16: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg17: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg18: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg19: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg20: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg21: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg22: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg23: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg24: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg25: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg26: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg27: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg28: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg29: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg30: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg31: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg32: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg33: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg34: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg35: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg36: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg37: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg38: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg39: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg40: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg41: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg42: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg43: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg44: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg45: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg46: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg47: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg48: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg49: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg50: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg51: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg52: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg53: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg54: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg55: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg56: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg57: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg58: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg59: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg60: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg61: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg62: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg63: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg64: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg65: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg66: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg67: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg68: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg69: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg70: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg71: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg72: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg73: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg74: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg75: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg76: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg77: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg78: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg79: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg80: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg81: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg82: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg83: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg84: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg85: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg86: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg87: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg88: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg89: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg90: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg91: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg92: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg93: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg94: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg95: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg96: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg97: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg98: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg99: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg100: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg101: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg102: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg103: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg104: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg105: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg106: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg107: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg108: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg109: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg110: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg111: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg112: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg113: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg114: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg115: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg116: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg117: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg118: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg119: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg120: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg121: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg122: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg123: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg124: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg125: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg126: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg127: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg128: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg129: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg130: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg131: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg132: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg133: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg134: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg135: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg136: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg137: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg138: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg139: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg140: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg141: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg142: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg143: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg144: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg145: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg146: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg147: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg148: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg149: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg150: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg151: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg152: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg153: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg154: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg155: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg156: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg157: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg158: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg159: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg160: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg161: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg162: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg163: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg164: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg165: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg166: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg167: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg168: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg169: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg170: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg171: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg172: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg173: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg174: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg175: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg176: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg177: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg178: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg179: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg180: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg181: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg182: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg183: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg184: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg185: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg186: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg187: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg188: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg189: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg190: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg191: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg192: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg193: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg194: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg195: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg196: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg197: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg198: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg199: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg200: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg201: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg202: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg203: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg204: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg205: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg206: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg207: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg208: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg209: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg210: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg211: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg212: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg213: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg214: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg215: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg216: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg217: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg218: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg219: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg220: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg221: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg222: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg223: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg224: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg225: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg226: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg227: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg228: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg229: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg230: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg231: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg232: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg233: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg234: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg235: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg236: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg237: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg238: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg239: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg240: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg241: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg242: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg243: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg244: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg245: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg246: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg247: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg248: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg249: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg250: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg251: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg252: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg253: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg254: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg255: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg256: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg257: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg258: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg259: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg260: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg261: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg262: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg263: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg264: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg265: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg266: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg267: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg268: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg269: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg270: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg271: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg272: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg273: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg274: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg275: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg276: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg277: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg278: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg279: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg280: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg281: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg282: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg283: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg284: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg285: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg286: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg287: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg288: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg289: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg290: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg291: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg292: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg293: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg294: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg295: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg296: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg297: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg298: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg299: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg300: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg301: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg302: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg303: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg304: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg305: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg306: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg307: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg308: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg309: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg310: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg311: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg312: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg313: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg314: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg315: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg316: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg317: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg318: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg319: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg320: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg321: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg322: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg323: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg324: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg325: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg326: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg327: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg328: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg329: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg330: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg331: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg332: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg333: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg334: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg335: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg336: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg337: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg338: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg339: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg340: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg341: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg342: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg343: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg344: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg345: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg346: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg347: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg348: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg349: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg350: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg351: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg352: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg353: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg354: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg355: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg356: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg357: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg358: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg359: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg360: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg361: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg362: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg363: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg364: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg365: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg366: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg367: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg368: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg369: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg370: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg371: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg372: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg373: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg374: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg375: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg376: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg377: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg378: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg379: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg380: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg381: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg382: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg383: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg384: !torch.tensor {torch.type_bound = !torch.vtensor<[4096],f32>}, %arg385: !torch.tensor {torch.type_bound = !torch.vtensor<[4096,1024],f32>}, %arg386: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg387: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg388: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg389: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,4096],f32>}, %arg390: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg391: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg392: !torch.tensor {torch.type_bound = !torch.vtensor<[50358],f32>}, %arg393: !torch.tensor {torch.type_bound = !torch.vtensor<[50358],f32>}, %arg394: !torch.tensor {torch.type_bound = !torch.vtensor<[50358,1024],f32>}, %arg395: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg396: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg397: !torch.tensor {torch.type_bound = !torch.vtensor<[1024],f32>}, %arg398: !torch.tensor {torch.type_bound = !torch.vtensor<[1024,1024],f32>}, %arg399: !torch.tensor {torch.type_bound = !torch.vtensor<[1,4096],si64>}, %arg400: !torch.tensor {torch.type_bound = !torch.vtensor<[1,4096],si64>}, %arg401: !torch.tensor {torch.type_bound = !torch.vtensor<[4,512],si64>}, %arg402: !torch.tensor {torch.type_bound = !torch.vtensor<[4,512],si64>}) -> !torch.list<!torch.tensor> {
    %none = torch.constant.none
    %1 = torch.tensor.literal(dense<1.000000e+00> : tensor<f64>) : !torch.tensor<[],f64>
    %2 = torch.tensor.literal(dense<0.79788456080286541> : tensor<f64>) : !torch.tensor<[],f64>
    %3 = torch.tensor.literal(dense<4.471500e-02> : tensor<f64>) : !torch.tensor<[],f64>
    %4 = torch.tensor.literal(dense<5.000000e-01> : tensor<f64>) : !torch.tensor<[],f64>
    %5 = torch.tensor.literal(dense<-0.000000e+00> : tensor<4x1x1x512xf32>) : !torch.tensor<[4,1,1,512],f32>
    %6 = torch.tensor.literal(dense<8.000000e+00> : tensor<f64>) : !torch.tensor<[],f64>
    %false = torch.constant.bool false
    %int0 = torch.constant.int 0
    %int9223372036854775807 = torch.constant.int 9223372036854775807
    %int1 = torch.constant.int 1
    %int512 = torch.constant.int 512
    %int4 = torch.constant.int 4
    %int1024 = torch.constant.int 1024
    %float9.999990e-13 = torch.constant.float 9.9999999999999998E-13
    %int2048 = torch.constant.int 2048
    %int16 = torch.constant.int 16
    %int64 = torch.constant.int 64
    %int2 = torch.constant.int 2
    %int3 = torch.constant.int 3
    %int-1 = torch.constant.int -1
    %int-2 = torch.constant.int -2
    %int4096 = torch.constant.int 4096
    %float3.000000e00 = torch.constant.float 3.000000e+00
    %int50358 = torch.constant.int 50358
    %int-100 = torch.constant.int -100
    %7 = torch.aten.slice.Tensor %arg400, %int0, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %8 = torch.aten.slice.Tensor %7, %int1, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %9 = torch.prim.ListConstruct %int4, %int512 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %10 = torch.aten.expand %8, %9, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %11 = torch.aten.slice.Tensor %arg399, %int0, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %12 = torch.aten.slice.Tensor %11, %int1, %int0, %int512, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %13 = torch.aten.embedding %arg5, %arg401, %int0, %false, %false : !torch.tensor, !torch.tensor, !torch.int, !torch.bool, !torch.bool -> !torch.tensor
    %14 = torch.aten.embedding %arg4, %10, %int-1, %false, %false : !torch.tensor, !torch.tensor, !torch.int, !torch.bool, !torch.bool -> !torch.tensor
    %15 = torch.aten.add.Tensor %13, %14, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %16 = torch.aten.embedding %arg3, %12, %int-1, %false, %false : !torch.tensor, !torch.tensor, !torch.int, !torch.bool, !torch.bool -> !torch.tensor
    %17 = torch.aten.add_.Tensor %15, %16, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %18 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0, %result1, %result2 = torch.aten.native_layer_norm %17, %18, %arg2, %arg1, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %19 = torch.aten.t %arg13 : !torch.tensor -> !torch.tensor
    %20 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %21 = torch.aten.view %result0, %20 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %22 = torch.aten.mm %21, %19 : !torch.tensor, !torch.tensor -> !torch.tensor
    %23 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %24 = torch.aten._unsafe_view %22, %23 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %25 = torch.aten.add_.Tensor %24, %arg12, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %26 = torch.aten.t %arg11 : !torch.tensor -> !torch.tensor
    %27 = torch.aten.mm %21, %26 : !torch.tensor, !torch.tensor -> !torch.tensor
    %28 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %29 = torch.aten._unsafe_view %27, %28 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %30 = torch.aten.add_.Tensor %29, %arg10, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %31 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %32 = torch.aten.view %30, %31 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %33 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %34 = torch.aten.permute %32, %33 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %35 = torch.aten.t %arg15 : !torch.tensor -> !torch.tensor
    %36 = torch.aten.mm %21, %35 : !torch.tensor, !torch.tensor -> !torch.tensor
    %37 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %38 = torch.aten._unsafe_view %36, %37 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %39 = torch.aten.add_.Tensor %38, %arg14, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %40 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %41 = torch.aten.view %39, %40 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %42 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %43 = torch.aten.permute %41, %42 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %44 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %45 = torch.aten.view %25, %44 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %46 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %47 = torch.aten.permute %45, %46 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %48 = torch.aten.transpose.int %34, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %49 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %50 = torch.aten.expand %47, %49, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %51 = torch.aten.clone %50, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %52 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %53 = torch.aten._unsafe_view %51, %52 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %54 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %55 = torch.aten.expand %48, %54, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %56 = torch.aten.clone %55, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %57 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %58 = torch.aten._unsafe_view %56, %57 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %59 = torch.aten.bmm %53, %58 : !torch.tensor, !torch.tensor -> !torch.tensor
    %60 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %61 = torch.aten._unsafe_view %59, %60 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %62 = torch.aten.div.Tensor %61, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %63 = torch.aten.add.Tensor %62, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %64 = torch.aten._softmax %63, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %65 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %66 = torch.aten.expand %64, %65, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %67 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %68 = torch.aten.view %66, %67 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %69 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %70 = torch.aten.expand %43, %69, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %71 = torch.aten.clone %70, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %72 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %73 = torch.aten._unsafe_view %71, %72 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %74 = torch.aten.bmm %68, %73 : !torch.tensor, !torch.tensor -> !torch.tensor
    %75 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %76 = torch.aten._unsafe_view %74, %75 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %77 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %78 = torch.aten.permute %76, %77 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %79 = torch.aten.clone %78, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %80 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %81 = torch.aten.view %79, %80 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %82 = torch.aten.t %arg9 : !torch.tensor -> !torch.tensor
    %83 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %84 = torch.aten.view %81, %83 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %85 = torch.aten.mm %84, %82 : !torch.tensor, !torch.tensor -> !torch.tensor
    %86 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %87 = torch.aten._unsafe_view %85, %86 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %88 = torch.aten.add_.Tensor %87, %arg8, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %89 = torch.aten.add.Tensor %88, %result0, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %90 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_0, %result1_1, %result2_2 = torch.aten.native_layer_norm %89, %90, %arg7, %arg6, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %91 = torch.aten.t %arg17 : !torch.tensor -> !torch.tensor
    %92 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %93 = torch.aten.view %result0_0, %92 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %94 = torch.aten.mm %93, %91 : !torch.tensor, !torch.tensor -> !torch.tensor
    %95 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %96 = torch.aten._unsafe_view %94, %95 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %97 = torch.aten.add_.Tensor %96, %arg16, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %98 = torch.aten.mul.Tensor %97, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %99 = torch.aten.pow.Tensor_Scalar %97, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %100 = torch.aten.mul.Tensor %99, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %101 = torch.aten.add.Tensor %97, %100, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %102 = torch.aten.mul.Tensor %101, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %103 = torch.aten.tanh %102 : !torch.tensor -> !torch.tensor
    %104 = torch.aten.add.Tensor %103, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %105 = torch.aten.mul.Tensor %98, %104 : !torch.tensor, !torch.tensor -> !torch.tensor
    %106 = torch.aten.t %arg21 : !torch.tensor -> !torch.tensor
    %107 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %108 = torch.aten.view %105, %107 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %109 = torch.aten.mm %108, %106 : !torch.tensor, !torch.tensor -> !torch.tensor
    %110 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %111 = torch.aten._unsafe_view %109, %110 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %112 = torch.aten.add_.Tensor %111, %arg20, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %113 = torch.aten.add.Tensor %112, %result0_0, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %114 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_3, %result1_4, %result2_5 = torch.aten.native_layer_norm %113, %114, %arg19, %arg18, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %115 = torch.aten.t %arg29 : !torch.tensor -> !torch.tensor
    %116 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %117 = torch.aten.view %result0_3, %116 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %118 = torch.aten.mm %117, %115 : !torch.tensor, !torch.tensor -> !torch.tensor
    %119 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %120 = torch.aten._unsafe_view %118, %119 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %121 = torch.aten.add_.Tensor %120, %arg28, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %122 = torch.aten.t %arg27 : !torch.tensor -> !torch.tensor
    %123 = torch.aten.mm %117, %122 : !torch.tensor, !torch.tensor -> !torch.tensor
    %124 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %125 = torch.aten._unsafe_view %123, %124 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %126 = torch.aten.add_.Tensor %125, %arg26, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %127 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %128 = torch.aten.view %126, %127 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %129 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %130 = torch.aten.permute %128, %129 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %131 = torch.aten.t %arg31 : !torch.tensor -> !torch.tensor
    %132 = torch.aten.mm %117, %131 : !torch.tensor, !torch.tensor -> !torch.tensor
    %133 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %134 = torch.aten._unsafe_view %132, %133 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %135 = torch.aten.add_.Tensor %134, %arg30, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %136 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %137 = torch.aten.view %135, %136 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %138 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %139 = torch.aten.permute %137, %138 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %140 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %141 = torch.aten.view %121, %140 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %142 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %143 = torch.aten.permute %141, %142 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %144 = torch.aten.transpose.int %130, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %145 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %146 = torch.aten.expand %143, %145, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %147 = torch.aten.clone %146, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %148 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %149 = torch.aten._unsafe_view %147, %148 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %150 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %151 = torch.aten.expand %144, %150, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %152 = torch.aten.clone %151, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %153 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %154 = torch.aten._unsafe_view %152, %153 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %155 = torch.aten.bmm %149, %154 : !torch.tensor, !torch.tensor -> !torch.tensor
    %156 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %157 = torch.aten._unsafe_view %155, %156 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %158 = torch.aten.div.Tensor %157, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %159 = torch.aten.add.Tensor %158, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %160 = torch.aten._softmax %159, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %161 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %162 = torch.aten.expand %160, %161, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %163 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %164 = torch.aten.view %162, %163 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %165 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %166 = torch.aten.expand %139, %165, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %167 = torch.aten.clone %166, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %168 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %169 = torch.aten._unsafe_view %167, %168 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %170 = torch.aten.bmm %164, %169 : !torch.tensor, !torch.tensor -> !torch.tensor
    %171 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %172 = torch.aten._unsafe_view %170, %171 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %173 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %174 = torch.aten.permute %172, %173 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %175 = torch.aten.clone %174, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %176 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %177 = torch.aten.view %175, %176 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %178 = torch.aten.t %arg25 : !torch.tensor -> !torch.tensor
    %179 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %180 = torch.aten.view %177, %179 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %181 = torch.aten.mm %180, %178 : !torch.tensor, !torch.tensor -> !torch.tensor
    %182 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %183 = torch.aten._unsafe_view %181, %182 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %184 = torch.aten.add_.Tensor %183, %arg24, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %185 = torch.aten.add.Tensor %184, %result0_3, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %186 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_6, %result1_7, %result2_8 = torch.aten.native_layer_norm %185, %186, %arg23, %arg22, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %187 = torch.aten.t %arg33 : !torch.tensor -> !torch.tensor
    %188 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %189 = torch.aten.view %result0_6, %188 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %190 = torch.aten.mm %189, %187 : !torch.tensor, !torch.tensor -> !torch.tensor
    %191 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %192 = torch.aten._unsafe_view %190, %191 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %193 = torch.aten.add_.Tensor %192, %arg32, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %194 = torch.aten.mul.Tensor %193, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %195 = torch.aten.pow.Tensor_Scalar %193, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %196 = torch.aten.mul.Tensor %195, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %197 = torch.aten.add.Tensor %193, %196, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %198 = torch.aten.mul.Tensor %197, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %199 = torch.aten.tanh %198 : !torch.tensor -> !torch.tensor
    %200 = torch.aten.add.Tensor %199, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %201 = torch.aten.mul.Tensor %194, %200 : !torch.tensor, !torch.tensor -> !torch.tensor
    %202 = torch.aten.t %arg37 : !torch.tensor -> !torch.tensor
    %203 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %204 = torch.aten.view %201, %203 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %205 = torch.aten.mm %204, %202 : !torch.tensor, !torch.tensor -> !torch.tensor
    %206 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %207 = torch.aten._unsafe_view %205, %206 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %208 = torch.aten.add_.Tensor %207, %arg36, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %209 = torch.aten.add.Tensor %208, %result0_6, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %210 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_9, %result1_10, %result2_11 = torch.aten.native_layer_norm %209, %210, %arg35, %arg34, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %211 = torch.aten.t %arg205 : !torch.tensor -> !torch.tensor
    %212 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %213 = torch.aten.view %result0_9, %212 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %214 = torch.aten.mm %213, %211 : !torch.tensor, !torch.tensor -> !torch.tensor
    %215 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %216 = torch.aten._unsafe_view %214, %215 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %217 = torch.aten.add_.Tensor %216, %arg204, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %218 = torch.aten.t %arg203 : !torch.tensor -> !torch.tensor
    %219 = torch.aten.mm %213, %218 : !torch.tensor, !torch.tensor -> !torch.tensor
    %220 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %221 = torch.aten._unsafe_view %219, %220 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %222 = torch.aten.add_.Tensor %221, %arg202, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %223 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %224 = torch.aten.view %222, %223 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %225 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %226 = torch.aten.permute %224, %225 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %227 = torch.aten.t %arg207 : !torch.tensor -> !torch.tensor
    %228 = torch.aten.mm %213, %227 : !torch.tensor, !torch.tensor -> !torch.tensor
    %229 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %230 = torch.aten._unsafe_view %228, %229 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %231 = torch.aten.add_.Tensor %230, %arg206, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %232 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %233 = torch.aten.view %231, %232 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %234 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %235 = torch.aten.permute %233, %234 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %236 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %237 = torch.aten.view %217, %236 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %238 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %239 = torch.aten.permute %237, %238 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %240 = torch.aten.transpose.int %226, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %241 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %242 = torch.aten.expand %239, %241, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %243 = torch.aten.clone %242, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %244 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %245 = torch.aten._unsafe_view %243, %244 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %246 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %247 = torch.aten.expand %240, %246, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %248 = torch.aten.clone %247, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %249 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %250 = torch.aten._unsafe_view %248, %249 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %251 = torch.aten.bmm %245, %250 : !torch.tensor, !torch.tensor -> !torch.tensor
    %252 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %253 = torch.aten._unsafe_view %251, %252 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %254 = torch.aten.div.Tensor %253, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %255 = torch.aten.add.Tensor %254, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %256 = torch.aten._softmax %255, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %257 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %258 = torch.aten.expand %256, %257, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %259 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %260 = torch.aten.view %258, %259 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %261 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %262 = torch.aten.expand %235, %261, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %263 = torch.aten.clone %262, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %264 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %265 = torch.aten._unsafe_view %263, %264 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %266 = torch.aten.bmm %260, %265 : !torch.tensor, !torch.tensor -> !torch.tensor
    %267 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %268 = torch.aten._unsafe_view %266, %267 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %269 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %270 = torch.aten.permute %268, %269 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %271 = torch.aten.clone %270, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %272 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %273 = torch.aten.view %271, %272 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %274 = torch.aten.t %arg201 : !torch.tensor -> !torch.tensor
    %275 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %276 = torch.aten.view %273, %275 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %277 = torch.aten.mm %276, %274 : !torch.tensor, !torch.tensor -> !torch.tensor
    %278 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %279 = torch.aten._unsafe_view %277, %278 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %280 = torch.aten.add_.Tensor %279, %arg200, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %281 = torch.aten.add.Tensor %280, %result0_9, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %282 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_12, %result1_13, %result2_14 = torch.aten.native_layer_norm %281, %282, %arg199, %arg198, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %283 = torch.aten.t %arg209 : !torch.tensor -> !torch.tensor
    %284 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %285 = torch.aten.view %result0_12, %284 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %286 = torch.aten.mm %285, %283 : !torch.tensor, !torch.tensor -> !torch.tensor
    %287 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %288 = torch.aten._unsafe_view %286, %287 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %289 = torch.aten.add_.Tensor %288, %arg208, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %290 = torch.aten.mul.Tensor %289, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %291 = torch.aten.pow.Tensor_Scalar %289, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %292 = torch.aten.mul.Tensor %291, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %293 = torch.aten.add.Tensor %289, %292, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %294 = torch.aten.mul.Tensor %293, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %295 = torch.aten.tanh %294 : !torch.tensor -> !torch.tensor
    %296 = torch.aten.add.Tensor %295, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %297 = torch.aten.mul.Tensor %290, %296 : !torch.tensor, !torch.tensor -> !torch.tensor
    %298 = torch.aten.t %arg213 : !torch.tensor -> !torch.tensor
    %299 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %300 = torch.aten.view %297, %299 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %301 = torch.aten.mm %300, %298 : !torch.tensor, !torch.tensor -> !torch.tensor
    %302 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %303 = torch.aten._unsafe_view %301, %302 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %304 = torch.aten.add_.Tensor %303, %arg212, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %305 = torch.aten.add.Tensor %304, %result0_12, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %306 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_15, %result1_16, %result2_17 = torch.aten.native_layer_norm %305, %306, %arg211, %arg210, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %307 = torch.aten.t %arg285 : !torch.tensor -> !torch.tensor
    %308 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %309 = torch.aten.view %result0_15, %308 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %310 = torch.aten.mm %309, %307 : !torch.tensor, !torch.tensor -> !torch.tensor
    %311 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %312 = torch.aten._unsafe_view %310, %311 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %313 = torch.aten.add_.Tensor %312, %arg284, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %314 = torch.aten.t %arg283 : !torch.tensor -> !torch.tensor
    %315 = torch.aten.mm %309, %314 : !torch.tensor, !torch.tensor -> !torch.tensor
    %316 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %317 = torch.aten._unsafe_view %315, %316 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %318 = torch.aten.add_.Tensor %317, %arg282, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %319 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %320 = torch.aten.view %318, %319 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %321 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %322 = torch.aten.permute %320, %321 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %323 = torch.aten.t %arg287 : !torch.tensor -> !torch.tensor
    %324 = torch.aten.mm %309, %323 : !torch.tensor, !torch.tensor -> !torch.tensor
    %325 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %326 = torch.aten._unsafe_view %324, %325 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %327 = torch.aten.add_.Tensor %326, %arg286, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %328 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %329 = torch.aten.view %327, %328 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %330 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %331 = torch.aten.permute %329, %330 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %332 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %333 = torch.aten.view %313, %332 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %334 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %335 = torch.aten.permute %333, %334 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %336 = torch.aten.transpose.int %322, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %337 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %338 = torch.aten.expand %335, %337, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %339 = torch.aten.clone %338, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %340 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %341 = torch.aten._unsafe_view %339, %340 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %342 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %343 = torch.aten.expand %336, %342, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %344 = torch.aten.clone %343, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %345 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %346 = torch.aten._unsafe_view %344, %345 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %347 = torch.aten.bmm %341, %346 : !torch.tensor, !torch.tensor -> !torch.tensor
    %348 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %349 = torch.aten._unsafe_view %347, %348 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %350 = torch.aten.div.Tensor %349, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %351 = torch.aten.add.Tensor %350, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %352 = torch.aten._softmax %351, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %353 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %354 = torch.aten.expand %352, %353, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %355 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %356 = torch.aten.view %354, %355 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %357 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %358 = torch.aten.expand %331, %357, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %359 = torch.aten.clone %358, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %360 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %361 = torch.aten._unsafe_view %359, %360 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %362 = torch.aten.bmm %356, %361 : !torch.tensor, !torch.tensor -> !torch.tensor
    %363 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %364 = torch.aten._unsafe_view %362, %363 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %365 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %366 = torch.aten.permute %364, %365 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %367 = torch.aten.clone %366, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %368 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %369 = torch.aten.view %367, %368 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %370 = torch.aten.t %arg281 : !torch.tensor -> !torch.tensor
    %371 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %372 = torch.aten.view %369, %371 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %373 = torch.aten.mm %372, %370 : !torch.tensor, !torch.tensor -> !torch.tensor
    %374 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %375 = torch.aten._unsafe_view %373, %374 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %376 = torch.aten.add_.Tensor %375, %arg280, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %377 = torch.aten.add.Tensor %376, %result0_15, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %378 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_18, %result1_19, %result2_20 = torch.aten.native_layer_norm %377, %378, %arg279, %arg278, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %379 = torch.aten.t %arg289 : !torch.tensor -> !torch.tensor
    %380 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %381 = torch.aten.view %result0_18, %380 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %382 = torch.aten.mm %381, %379 : !torch.tensor, !torch.tensor -> !torch.tensor
    %383 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %384 = torch.aten._unsafe_view %382, %383 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %385 = torch.aten.add_.Tensor %384, %arg288, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %386 = torch.aten.mul.Tensor %385, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %387 = torch.aten.pow.Tensor_Scalar %385, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %388 = torch.aten.mul.Tensor %387, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %389 = torch.aten.add.Tensor %385, %388, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %390 = torch.aten.mul.Tensor %389, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %391 = torch.aten.tanh %390 : !torch.tensor -> !torch.tensor
    %392 = torch.aten.add.Tensor %391, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %393 = torch.aten.mul.Tensor %386, %392 : !torch.tensor, !torch.tensor -> !torch.tensor
    %394 = torch.aten.t %arg293 : !torch.tensor -> !torch.tensor
    %395 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %396 = torch.aten.view %393, %395 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %397 = torch.aten.mm %396, %394 : !torch.tensor, !torch.tensor -> !torch.tensor
    %398 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %399 = torch.aten._unsafe_view %397, %398 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %400 = torch.aten.add_.Tensor %399, %arg292, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %401 = torch.aten.add.Tensor %400, %result0_18, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %402 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_21, %result1_22, %result2_23 = torch.aten.native_layer_norm %401, %402, %arg291, %arg290, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %403 = torch.aten.t %arg301 : !torch.tensor -> !torch.tensor
    %404 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %405 = torch.aten.view %result0_21, %404 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %406 = torch.aten.mm %405, %403 : !torch.tensor, !torch.tensor -> !torch.tensor
    %407 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %408 = torch.aten._unsafe_view %406, %407 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %409 = torch.aten.add_.Tensor %408, %arg300, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %410 = torch.aten.t %arg299 : !torch.tensor -> !torch.tensor
    %411 = torch.aten.mm %405, %410 : !torch.tensor, !torch.tensor -> !torch.tensor
    %412 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %413 = torch.aten._unsafe_view %411, %412 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %414 = torch.aten.add_.Tensor %413, %arg298, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %415 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %416 = torch.aten.view %414, %415 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %417 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %418 = torch.aten.permute %416, %417 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %419 = torch.aten.t %arg303 : !torch.tensor -> !torch.tensor
    %420 = torch.aten.mm %405, %419 : !torch.tensor, !torch.tensor -> !torch.tensor
    %421 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %422 = torch.aten._unsafe_view %420, %421 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %423 = torch.aten.add_.Tensor %422, %arg302, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %424 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %425 = torch.aten.view %423, %424 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %426 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %427 = torch.aten.permute %425, %426 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %428 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %429 = torch.aten.view %409, %428 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %430 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %431 = torch.aten.permute %429, %430 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %432 = torch.aten.transpose.int %418, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %433 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %434 = torch.aten.expand %431, %433, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %435 = torch.aten.clone %434, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %436 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %437 = torch.aten._unsafe_view %435, %436 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %438 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %439 = torch.aten.expand %432, %438, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %440 = torch.aten.clone %439, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %441 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %442 = torch.aten._unsafe_view %440, %441 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %443 = torch.aten.bmm %437, %442 : !torch.tensor, !torch.tensor -> !torch.tensor
    %444 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %445 = torch.aten._unsafe_view %443, %444 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %446 = torch.aten.div.Tensor %445, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %447 = torch.aten.add.Tensor %446, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %448 = torch.aten._softmax %447, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %449 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %450 = torch.aten.expand %448, %449, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %451 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %452 = torch.aten.view %450, %451 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %453 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %454 = torch.aten.expand %427, %453, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %455 = torch.aten.clone %454, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %456 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %457 = torch.aten._unsafe_view %455, %456 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %458 = torch.aten.bmm %452, %457 : !torch.tensor, !torch.tensor -> !torch.tensor
    %459 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %460 = torch.aten._unsafe_view %458, %459 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %461 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %462 = torch.aten.permute %460, %461 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %463 = torch.aten.clone %462, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %464 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %465 = torch.aten.view %463, %464 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %466 = torch.aten.t %arg297 : !torch.tensor -> !torch.tensor
    %467 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %468 = torch.aten.view %465, %467 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %469 = torch.aten.mm %468, %466 : !torch.tensor, !torch.tensor -> !torch.tensor
    %470 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %471 = torch.aten._unsafe_view %469, %470 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %472 = torch.aten.add_.Tensor %471, %arg296, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %473 = torch.aten.add.Tensor %472, %result0_21, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %474 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_24, %result1_25, %result2_26 = torch.aten.native_layer_norm %473, %474, %arg295, %arg294, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %475 = torch.aten.t %arg305 : !torch.tensor -> !torch.tensor
    %476 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %477 = torch.aten.view %result0_24, %476 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %478 = torch.aten.mm %477, %475 : !torch.tensor, !torch.tensor -> !torch.tensor
    %479 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %480 = torch.aten._unsafe_view %478, %479 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %481 = torch.aten.add_.Tensor %480, %arg304, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %482 = torch.aten.mul.Tensor %481, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %483 = torch.aten.pow.Tensor_Scalar %481, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %484 = torch.aten.mul.Tensor %483, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %485 = torch.aten.add.Tensor %481, %484, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %486 = torch.aten.mul.Tensor %485, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %487 = torch.aten.tanh %486 : !torch.tensor -> !torch.tensor
    %488 = torch.aten.add.Tensor %487, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %489 = torch.aten.mul.Tensor %482, %488 : !torch.tensor, !torch.tensor -> !torch.tensor
    %490 = torch.aten.t %arg309 : !torch.tensor -> !torch.tensor
    %491 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %492 = torch.aten.view %489, %491 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %493 = torch.aten.mm %492, %490 : !torch.tensor, !torch.tensor -> !torch.tensor
    %494 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %495 = torch.aten._unsafe_view %493, %494 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %496 = torch.aten.add_.Tensor %495, %arg308, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %497 = torch.aten.add.Tensor %496, %result0_24, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %498 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_27, %result1_28, %result2_29 = torch.aten.native_layer_norm %497, %498, %arg307, %arg306, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %499 = torch.aten.t %arg317 : !torch.tensor -> !torch.tensor
    %500 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %501 = torch.aten.view %result0_27, %500 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %502 = torch.aten.mm %501, %499 : !torch.tensor, !torch.tensor -> !torch.tensor
    %503 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %504 = torch.aten._unsafe_view %502, %503 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %505 = torch.aten.add_.Tensor %504, %arg316, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %506 = torch.aten.t %arg315 : !torch.tensor -> !torch.tensor
    %507 = torch.aten.mm %501, %506 : !torch.tensor, !torch.tensor -> !torch.tensor
    %508 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %509 = torch.aten._unsafe_view %507, %508 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %510 = torch.aten.add_.Tensor %509, %arg314, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %511 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %512 = torch.aten.view %510, %511 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %513 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %514 = torch.aten.permute %512, %513 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %515 = torch.aten.t %arg319 : !torch.tensor -> !torch.tensor
    %516 = torch.aten.mm %501, %515 : !torch.tensor, !torch.tensor -> !torch.tensor
    %517 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %518 = torch.aten._unsafe_view %516, %517 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %519 = torch.aten.add_.Tensor %518, %arg318, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %520 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %521 = torch.aten.view %519, %520 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %522 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %523 = torch.aten.permute %521, %522 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %524 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %525 = torch.aten.view %505, %524 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %526 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %527 = torch.aten.permute %525, %526 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %528 = torch.aten.transpose.int %514, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %529 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %530 = torch.aten.expand %527, %529, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %531 = torch.aten.clone %530, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %532 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %533 = torch.aten._unsafe_view %531, %532 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %534 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %535 = torch.aten.expand %528, %534, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %536 = torch.aten.clone %535, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %537 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %538 = torch.aten._unsafe_view %536, %537 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %539 = torch.aten.bmm %533, %538 : !torch.tensor, !torch.tensor -> !torch.tensor
    %540 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %541 = torch.aten._unsafe_view %539, %540 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %542 = torch.aten.div.Tensor %541, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %543 = torch.aten.add.Tensor %542, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %544 = torch.aten._softmax %543, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %545 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %546 = torch.aten.expand %544, %545, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %547 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %548 = torch.aten.view %546, %547 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %549 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %550 = torch.aten.expand %523, %549, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %551 = torch.aten.clone %550, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %552 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %553 = torch.aten._unsafe_view %551, %552 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %554 = torch.aten.bmm %548, %553 : !torch.tensor, !torch.tensor -> !torch.tensor
    %555 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %556 = torch.aten._unsafe_view %554, %555 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %557 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %558 = torch.aten.permute %556, %557 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %559 = torch.aten.clone %558, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %560 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %561 = torch.aten.view %559, %560 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %562 = torch.aten.t %arg313 : !torch.tensor -> !torch.tensor
    %563 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %564 = torch.aten.view %561, %563 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %565 = torch.aten.mm %564, %562 : !torch.tensor, !torch.tensor -> !torch.tensor
    %566 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %567 = torch.aten._unsafe_view %565, %566 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %568 = torch.aten.add_.Tensor %567, %arg312, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %569 = torch.aten.add.Tensor %568, %result0_27, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %570 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_30, %result1_31, %result2_32 = torch.aten.native_layer_norm %569, %570, %arg311, %arg310, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %571 = torch.aten.t %arg321 : !torch.tensor -> !torch.tensor
    %572 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %573 = torch.aten.view %result0_30, %572 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %574 = torch.aten.mm %573, %571 : !torch.tensor, !torch.tensor -> !torch.tensor
    %575 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %576 = torch.aten._unsafe_view %574, %575 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %577 = torch.aten.add_.Tensor %576, %arg320, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %578 = torch.aten.mul.Tensor %577, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %579 = torch.aten.pow.Tensor_Scalar %577, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %580 = torch.aten.mul.Tensor %579, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %581 = torch.aten.add.Tensor %577, %580, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %582 = torch.aten.mul.Tensor %581, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %583 = torch.aten.tanh %582 : !torch.tensor -> !torch.tensor
    %584 = torch.aten.add.Tensor %583, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %585 = torch.aten.mul.Tensor %578, %584 : !torch.tensor, !torch.tensor -> !torch.tensor
    %586 = torch.aten.t %arg325 : !torch.tensor -> !torch.tensor
    %587 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %588 = torch.aten.view %585, %587 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %589 = torch.aten.mm %588, %586 : !torch.tensor, !torch.tensor -> !torch.tensor
    %590 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %591 = torch.aten._unsafe_view %589, %590 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %592 = torch.aten.add_.Tensor %591, %arg324, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %593 = torch.aten.add.Tensor %592, %result0_30, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %594 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_33, %result1_34, %result2_35 = torch.aten.native_layer_norm %593, %594, %arg323, %arg322, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %595 = torch.aten.t %arg333 : !torch.tensor -> !torch.tensor
    %596 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %597 = torch.aten.view %result0_33, %596 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %598 = torch.aten.mm %597, %595 : !torch.tensor, !torch.tensor -> !torch.tensor
    %599 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %600 = torch.aten._unsafe_view %598, %599 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %601 = torch.aten.add_.Tensor %600, %arg332, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %602 = torch.aten.t %arg331 : !torch.tensor -> !torch.tensor
    %603 = torch.aten.mm %597, %602 : !torch.tensor, !torch.tensor -> !torch.tensor
    %604 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %605 = torch.aten._unsafe_view %603, %604 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %606 = torch.aten.add_.Tensor %605, %arg330, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %607 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %608 = torch.aten.view %606, %607 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %609 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %610 = torch.aten.permute %608, %609 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %611 = torch.aten.t %arg335 : !torch.tensor -> !torch.tensor
    %612 = torch.aten.mm %597, %611 : !torch.tensor, !torch.tensor -> !torch.tensor
    %613 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %614 = torch.aten._unsafe_view %612, %613 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %615 = torch.aten.add_.Tensor %614, %arg334, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %616 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %617 = torch.aten.view %615, %616 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %618 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %619 = torch.aten.permute %617, %618 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %620 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %621 = torch.aten.view %601, %620 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %622 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %623 = torch.aten.permute %621, %622 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %624 = torch.aten.transpose.int %610, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %625 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %626 = torch.aten.expand %623, %625, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %627 = torch.aten.clone %626, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %628 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %629 = torch.aten._unsafe_view %627, %628 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %630 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %631 = torch.aten.expand %624, %630, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %632 = torch.aten.clone %631, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %633 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %634 = torch.aten._unsafe_view %632, %633 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %635 = torch.aten.bmm %629, %634 : !torch.tensor, !torch.tensor -> !torch.tensor
    %636 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %637 = torch.aten._unsafe_view %635, %636 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %638 = torch.aten.div.Tensor %637, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %639 = torch.aten.add.Tensor %638, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %640 = torch.aten._softmax %639, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %641 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %642 = torch.aten.expand %640, %641, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %643 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %644 = torch.aten.view %642, %643 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %645 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %646 = torch.aten.expand %619, %645, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %647 = torch.aten.clone %646, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %648 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %649 = torch.aten._unsafe_view %647, %648 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %650 = torch.aten.bmm %644, %649 : !torch.tensor, !torch.tensor -> !torch.tensor
    %651 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %652 = torch.aten._unsafe_view %650, %651 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %653 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %654 = torch.aten.permute %652, %653 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %655 = torch.aten.clone %654, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %656 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %657 = torch.aten.view %655, %656 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %658 = torch.aten.t %arg329 : !torch.tensor -> !torch.tensor
    %659 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %660 = torch.aten.view %657, %659 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %661 = torch.aten.mm %660, %658 : !torch.tensor, !torch.tensor -> !torch.tensor
    %662 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %663 = torch.aten._unsafe_view %661, %662 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %664 = torch.aten.add_.Tensor %663, %arg328, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %665 = torch.aten.add.Tensor %664, %result0_33, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %666 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_36, %result1_37, %result2_38 = torch.aten.native_layer_norm %665, %666, %arg327, %arg326, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %667 = torch.aten.t %arg337 : !torch.tensor -> !torch.tensor
    %668 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %669 = torch.aten.view %result0_36, %668 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %670 = torch.aten.mm %669, %667 : !torch.tensor, !torch.tensor -> !torch.tensor
    %671 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %672 = torch.aten._unsafe_view %670, %671 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %673 = torch.aten.add_.Tensor %672, %arg336, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %674 = torch.aten.mul.Tensor %673, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %675 = torch.aten.pow.Tensor_Scalar %673, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %676 = torch.aten.mul.Tensor %675, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %677 = torch.aten.add.Tensor %673, %676, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %678 = torch.aten.mul.Tensor %677, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %679 = torch.aten.tanh %678 : !torch.tensor -> !torch.tensor
    %680 = torch.aten.add.Tensor %679, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %681 = torch.aten.mul.Tensor %674, %680 : !torch.tensor, !torch.tensor -> !torch.tensor
    %682 = torch.aten.t %arg341 : !torch.tensor -> !torch.tensor
    %683 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %684 = torch.aten.view %681, %683 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %685 = torch.aten.mm %684, %682 : !torch.tensor, !torch.tensor -> !torch.tensor
    %686 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %687 = torch.aten._unsafe_view %685, %686 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %688 = torch.aten.add_.Tensor %687, %arg340, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %689 = torch.aten.add.Tensor %688, %result0_36, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %690 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_39, %result1_40, %result2_41 = torch.aten.native_layer_norm %689, %690, %arg339, %arg338, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %691 = torch.aten.t %arg349 : !torch.tensor -> !torch.tensor
    %692 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %693 = torch.aten.view %result0_39, %692 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %694 = torch.aten.mm %693, %691 : !torch.tensor, !torch.tensor -> !torch.tensor
    %695 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %696 = torch.aten._unsafe_view %694, %695 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %697 = torch.aten.add_.Tensor %696, %arg348, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %698 = torch.aten.t %arg347 : !torch.tensor -> !torch.tensor
    %699 = torch.aten.mm %693, %698 : !torch.tensor, !torch.tensor -> !torch.tensor
    %700 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %701 = torch.aten._unsafe_view %699, %700 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %702 = torch.aten.add_.Tensor %701, %arg346, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %703 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %704 = torch.aten.view %702, %703 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %705 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %706 = torch.aten.permute %704, %705 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %707 = torch.aten.t %arg351 : !torch.tensor -> !torch.tensor
    %708 = torch.aten.mm %693, %707 : !torch.tensor, !torch.tensor -> !torch.tensor
    %709 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %710 = torch.aten._unsafe_view %708, %709 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %711 = torch.aten.add_.Tensor %710, %arg350, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %712 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %713 = torch.aten.view %711, %712 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %714 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %715 = torch.aten.permute %713, %714 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %716 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %717 = torch.aten.view %697, %716 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %718 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %719 = torch.aten.permute %717, %718 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %720 = torch.aten.transpose.int %706, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %721 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %722 = torch.aten.expand %719, %721, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %723 = torch.aten.clone %722, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %724 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %725 = torch.aten._unsafe_view %723, %724 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %726 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %727 = torch.aten.expand %720, %726, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %728 = torch.aten.clone %727, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %729 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %730 = torch.aten._unsafe_view %728, %729 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %731 = torch.aten.bmm %725, %730 : !torch.tensor, !torch.tensor -> !torch.tensor
    %732 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %733 = torch.aten._unsafe_view %731, %732 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %734 = torch.aten.div.Tensor %733, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %735 = torch.aten.add.Tensor %734, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %736 = torch.aten._softmax %735, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %737 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %738 = torch.aten.expand %736, %737, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %739 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %740 = torch.aten.view %738, %739 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %741 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %742 = torch.aten.expand %715, %741, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %743 = torch.aten.clone %742, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %744 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %745 = torch.aten._unsafe_view %743, %744 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %746 = torch.aten.bmm %740, %745 : !torch.tensor, !torch.tensor -> !torch.tensor
    %747 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %748 = torch.aten._unsafe_view %746, %747 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %749 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %750 = torch.aten.permute %748, %749 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %751 = torch.aten.clone %750, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %752 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %753 = torch.aten.view %751, %752 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %754 = torch.aten.t %arg345 : !torch.tensor -> !torch.tensor
    %755 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %756 = torch.aten.view %753, %755 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %757 = torch.aten.mm %756, %754 : !torch.tensor, !torch.tensor -> !torch.tensor
    %758 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %759 = torch.aten._unsafe_view %757, %758 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %760 = torch.aten.add_.Tensor %759, %arg344, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %761 = torch.aten.add.Tensor %760, %result0_39, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %762 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_42, %result1_43, %result2_44 = torch.aten.native_layer_norm %761, %762, %arg343, %arg342, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %763 = torch.aten.t %arg353 : !torch.tensor -> !torch.tensor
    %764 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %765 = torch.aten.view %result0_42, %764 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %766 = torch.aten.mm %765, %763 : !torch.tensor, !torch.tensor -> !torch.tensor
    %767 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %768 = torch.aten._unsafe_view %766, %767 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %769 = torch.aten.add_.Tensor %768, %arg352, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %770 = torch.aten.mul.Tensor %769, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %771 = torch.aten.pow.Tensor_Scalar %769, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %772 = torch.aten.mul.Tensor %771, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %773 = torch.aten.add.Tensor %769, %772, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %774 = torch.aten.mul.Tensor %773, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %775 = torch.aten.tanh %774 : !torch.tensor -> !torch.tensor
    %776 = torch.aten.add.Tensor %775, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %777 = torch.aten.mul.Tensor %770, %776 : !torch.tensor, !torch.tensor -> !torch.tensor
    %778 = torch.aten.t %arg357 : !torch.tensor -> !torch.tensor
    %779 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %780 = torch.aten.view %777, %779 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %781 = torch.aten.mm %780, %778 : !torch.tensor, !torch.tensor -> !torch.tensor
    %782 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %783 = torch.aten._unsafe_view %781, %782 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %784 = torch.aten.add_.Tensor %783, %arg356, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %785 = torch.aten.add.Tensor %784, %result0_42, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %786 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_45, %result1_46, %result2_47 = torch.aten.native_layer_norm %785, %786, %arg355, %arg354, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %787 = torch.aten.t %arg365 : !torch.tensor -> !torch.tensor
    %788 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %789 = torch.aten.view %result0_45, %788 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %790 = torch.aten.mm %789, %787 : !torch.tensor, !torch.tensor -> !torch.tensor
    %791 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %792 = torch.aten._unsafe_view %790, %791 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %793 = torch.aten.add_.Tensor %792, %arg364, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %794 = torch.aten.t %arg363 : !torch.tensor -> !torch.tensor
    %795 = torch.aten.mm %789, %794 : !torch.tensor, !torch.tensor -> !torch.tensor
    %796 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %797 = torch.aten._unsafe_view %795, %796 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %798 = torch.aten.add_.Tensor %797, %arg362, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %799 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %800 = torch.aten.view %798, %799 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %801 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %802 = torch.aten.permute %800, %801 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %803 = torch.aten.t %arg367 : !torch.tensor -> !torch.tensor
    %804 = torch.aten.mm %789, %803 : !torch.tensor, !torch.tensor -> !torch.tensor
    %805 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %806 = torch.aten._unsafe_view %804, %805 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %807 = torch.aten.add_.Tensor %806, %arg366, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %808 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %809 = torch.aten.view %807, %808 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %810 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %811 = torch.aten.permute %809, %810 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %812 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %813 = torch.aten.view %793, %812 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %814 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %815 = torch.aten.permute %813, %814 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %816 = torch.aten.transpose.int %802, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %817 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %818 = torch.aten.expand %815, %817, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %819 = torch.aten.clone %818, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %820 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %821 = torch.aten._unsafe_view %819, %820 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %822 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %823 = torch.aten.expand %816, %822, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %824 = torch.aten.clone %823, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %825 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %826 = torch.aten._unsafe_view %824, %825 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %827 = torch.aten.bmm %821, %826 : !torch.tensor, !torch.tensor -> !torch.tensor
    %828 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %829 = torch.aten._unsafe_view %827, %828 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %830 = torch.aten.div.Tensor %829, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %831 = torch.aten.add.Tensor %830, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %832 = torch.aten._softmax %831, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %833 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %834 = torch.aten.expand %832, %833, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %835 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %836 = torch.aten.view %834, %835 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %837 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %838 = torch.aten.expand %811, %837, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %839 = torch.aten.clone %838, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %840 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %841 = torch.aten._unsafe_view %839, %840 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %842 = torch.aten.bmm %836, %841 : !torch.tensor, !torch.tensor -> !torch.tensor
    %843 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %844 = torch.aten._unsafe_view %842, %843 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %845 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %846 = torch.aten.permute %844, %845 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %847 = torch.aten.clone %846, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %848 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %849 = torch.aten.view %847, %848 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %850 = torch.aten.t %arg361 : !torch.tensor -> !torch.tensor
    %851 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %852 = torch.aten.view %849, %851 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %853 = torch.aten.mm %852, %850 : !torch.tensor, !torch.tensor -> !torch.tensor
    %854 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %855 = torch.aten._unsafe_view %853, %854 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %856 = torch.aten.add_.Tensor %855, %arg360, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %857 = torch.aten.add.Tensor %856, %result0_45, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %858 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_48, %result1_49, %result2_50 = torch.aten.native_layer_norm %857, %858, %arg359, %arg358, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %859 = torch.aten.t %arg369 : !torch.tensor -> !torch.tensor
    %860 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %861 = torch.aten.view %result0_48, %860 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %862 = torch.aten.mm %861, %859 : !torch.tensor, !torch.tensor -> !torch.tensor
    %863 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %864 = torch.aten._unsafe_view %862, %863 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %865 = torch.aten.add_.Tensor %864, %arg368, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %866 = torch.aten.mul.Tensor %865, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %867 = torch.aten.pow.Tensor_Scalar %865, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %868 = torch.aten.mul.Tensor %867, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %869 = torch.aten.add.Tensor %865, %868, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %870 = torch.aten.mul.Tensor %869, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %871 = torch.aten.tanh %870 : !torch.tensor -> !torch.tensor
    %872 = torch.aten.add.Tensor %871, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %873 = torch.aten.mul.Tensor %866, %872 : !torch.tensor, !torch.tensor -> !torch.tensor
    %874 = torch.aten.t %arg373 : !torch.tensor -> !torch.tensor
    %875 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %876 = torch.aten.view %873, %875 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %877 = torch.aten.mm %876, %874 : !torch.tensor, !torch.tensor -> !torch.tensor
    %878 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %879 = torch.aten._unsafe_view %877, %878 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %880 = torch.aten.add_.Tensor %879, %arg372, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %881 = torch.aten.add.Tensor %880, %result0_48, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %882 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_51, %result1_52, %result2_53 = torch.aten.native_layer_norm %881, %882, %arg371, %arg370, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %883 = torch.aten.t %arg381 : !torch.tensor -> !torch.tensor
    %884 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %885 = torch.aten.view %result0_51, %884 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %886 = torch.aten.mm %885, %883 : !torch.tensor, !torch.tensor -> !torch.tensor
    %887 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %888 = torch.aten._unsafe_view %886, %887 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %889 = torch.aten.add_.Tensor %888, %arg380, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %890 = torch.aten.t %arg379 : !torch.tensor -> !torch.tensor
    %891 = torch.aten.mm %885, %890 : !torch.tensor, !torch.tensor -> !torch.tensor
    %892 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %893 = torch.aten._unsafe_view %891, %892 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %894 = torch.aten.add_.Tensor %893, %arg378, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %895 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %896 = torch.aten.view %894, %895 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %897 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %898 = torch.aten.permute %896, %897 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %899 = torch.aten.t %arg383 : !torch.tensor -> !torch.tensor
    %900 = torch.aten.mm %885, %899 : !torch.tensor, !torch.tensor -> !torch.tensor
    %901 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %902 = torch.aten._unsafe_view %900, %901 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %903 = torch.aten.add_.Tensor %902, %arg382, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %904 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %905 = torch.aten.view %903, %904 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %906 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %907 = torch.aten.permute %905, %906 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %908 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %909 = torch.aten.view %889, %908 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %910 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %911 = torch.aten.permute %909, %910 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %912 = torch.aten.transpose.int %898, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %913 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %914 = torch.aten.expand %911, %913, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %915 = torch.aten.clone %914, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %916 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %917 = torch.aten._unsafe_view %915, %916 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %918 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %919 = torch.aten.expand %912, %918, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %920 = torch.aten.clone %919, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %921 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %922 = torch.aten._unsafe_view %920, %921 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %923 = torch.aten.bmm %917, %922 : !torch.tensor, !torch.tensor -> !torch.tensor
    %924 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %925 = torch.aten._unsafe_view %923, %924 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %926 = torch.aten.div.Tensor %925, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %927 = torch.aten.add.Tensor %926, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %928 = torch.aten._softmax %927, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %929 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %930 = torch.aten.expand %928, %929, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %931 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %932 = torch.aten.view %930, %931 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %933 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %934 = torch.aten.expand %907, %933, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %935 = torch.aten.clone %934, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %936 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %937 = torch.aten._unsafe_view %935, %936 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %938 = torch.aten.bmm %932, %937 : !torch.tensor, !torch.tensor -> !torch.tensor
    %939 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %940 = torch.aten._unsafe_view %938, %939 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %941 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %942 = torch.aten.permute %940, %941 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %943 = torch.aten.clone %942, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %944 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %945 = torch.aten.view %943, %944 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %946 = torch.aten.t %arg377 : !torch.tensor -> !torch.tensor
    %947 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %948 = torch.aten.view %945, %947 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %949 = torch.aten.mm %948, %946 : !torch.tensor, !torch.tensor -> !torch.tensor
    %950 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %951 = torch.aten._unsafe_view %949, %950 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %952 = torch.aten.add_.Tensor %951, %arg376, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %953 = torch.aten.add.Tensor %952, %result0_51, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %954 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_54, %result1_55, %result2_56 = torch.aten.native_layer_norm %953, %954, %arg375, %arg374, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %955 = torch.aten.t %arg385 : !torch.tensor -> !torch.tensor
    %956 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %957 = torch.aten.view %result0_54, %956 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %958 = torch.aten.mm %957, %955 : !torch.tensor, !torch.tensor -> !torch.tensor
    %959 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %960 = torch.aten._unsafe_view %958, %959 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %961 = torch.aten.add_.Tensor %960, %arg384, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %962 = torch.aten.mul.Tensor %961, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %963 = torch.aten.pow.Tensor_Scalar %961, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %964 = torch.aten.mul.Tensor %963, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %965 = torch.aten.add.Tensor %961, %964, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %966 = torch.aten.mul.Tensor %965, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %967 = torch.aten.tanh %966 : !torch.tensor -> !torch.tensor
    %968 = torch.aten.add.Tensor %967, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %969 = torch.aten.mul.Tensor %962, %968 : !torch.tensor, !torch.tensor -> !torch.tensor
    %970 = torch.aten.t %arg389 : !torch.tensor -> !torch.tensor
    %971 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %972 = torch.aten.view %969, %971 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %973 = torch.aten.mm %972, %970 : !torch.tensor, !torch.tensor -> !torch.tensor
    %974 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %975 = torch.aten._unsafe_view %973, %974 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %976 = torch.aten.add_.Tensor %975, %arg388, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %977 = torch.aten.add.Tensor %976, %result0_54, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %978 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_57, %result1_58, %result2_59 = torch.aten.native_layer_norm %977, %978, %arg387, %arg386, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %979 = torch.aten.t %arg45 : !torch.tensor -> !torch.tensor
    %980 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %981 = torch.aten.view %result0_57, %980 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %982 = torch.aten.mm %981, %979 : !torch.tensor, !torch.tensor -> !torch.tensor
    %983 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %984 = torch.aten._unsafe_view %982, %983 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %985 = torch.aten.add_.Tensor %984, %arg44, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %986 = torch.aten.t %arg43 : !torch.tensor -> !torch.tensor
    %987 = torch.aten.mm %981, %986 : !torch.tensor, !torch.tensor -> !torch.tensor
    %988 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %989 = torch.aten._unsafe_view %987, %988 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %990 = torch.aten.add_.Tensor %989, %arg42, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %991 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %992 = torch.aten.view %990, %991 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %993 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %994 = torch.aten.permute %992, %993 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %995 = torch.aten.t %arg47 : !torch.tensor -> !torch.tensor
    %996 = torch.aten.mm %981, %995 : !torch.tensor, !torch.tensor -> !torch.tensor
    %997 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %998 = torch.aten._unsafe_view %996, %997 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %999 = torch.aten.add_.Tensor %998, %arg46, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1000 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1001 = torch.aten.view %999, %1000 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1002 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1003 = torch.aten.permute %1001, %1002 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1004 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1005 = torch.aten.view %985, %1004 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1006 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1007 = torch.aten.permute %1005, %1006 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1008 = torch.aten.transpose.int %994, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1009 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1010 = torch.aten.expand %1007, %1009, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1011 = torch.aten.clone %1010, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1012 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1013 = torch.aten._unsafe_view %1011, %1012 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1014 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1015 = torch.aten.expand %1008, %1014, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1016 = torch.aten.clone %1015, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1017 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1018 = torch.aten._unsafe_view %1016, %1017 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1019 = torch.aten.bmm %1013, %1018 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1020 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1021 = torch.aten._unsafe_view %1019, %1020 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1022 = torch.aten.div.Tensor %1021, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1023 = torch.aten.add.Tensor %1022, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %1024 = torch.aten._softmax %1023, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %1025 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1026 = torch.aten.expand %1024, %1025, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1027 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1028 = torch.aten.view %1026, %1027 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1029 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1030 = torch.aten.expand %1003, %1029, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1031 = torch.aten.clone %1030, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1032 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1033 = torch.aten._unsafe_view %1031, %1032 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1034 = torch.aten.bmm %1028, %1033 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1035 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1036 = torch.aten._unsafe_view %1034, %1035 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1037 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1038 = torch.aten.permute %1036, %1037 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1039 = torch.aten.clone %1038, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1040 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1041 = torch.aten.view %1039, %1040 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1042 = torch.aten.t %arg41 : !torch.tensor -> !torch.tensor
    %1043 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1044 = torch.aten.view %1041, %1043 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1045 = torch.aten.mm %1044, %1042 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1046 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1047 = torch.aten._unsafe_view %1045, %1046 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1048 = torch.aten.add_.Tensor %1047, %arg40, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1049 = torch.aten.add.Tensor %1048, %result0_57, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1050 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_60, %result1_61, %result2_62 = torch.aten.native_layer_norm %1049, %1050, %arg39, %arg38, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1051 = torch.aten.t %arg49 : !torch.tensor -> !torch.tensor
    %1052 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1053 = torch.aten.view %result0_60, %1052 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1054 = torch.aten.mm %1053, %1051 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1055 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1056 = torch.aten._unsafe_view %1054, %1055 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1057 = torch.aten.add_.Tensor %1056, %arg48, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1058 = torch.aten.mul.Tensor %1057, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1059 = torch.aten.pow.Tensor_Scalar %1057, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %1060 = torch.aten.mul.Tensor %1059, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1061 = torch.aten.add.Tensor %1057, %1060, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1062 = torch.aten.mul.Tensor %1061, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1063 = torch.aten.tanh %1062 : !torch.tensor -> !torch.tensor
    %1064 = torch.aten.add.Tensor %1063, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %1065 = torch.aten.mul.Tensor %1058, %1064 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1066 = torch.aten.t %arg53 : !torch.tensor -> !torch.tensor
    %1067 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1068 = torch.aten.view %1065, %1067 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1069 = torch.aten.mm %1068, %1066 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1070 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1071 = torch.aten._unsafe_view %1069, %1070 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1072 = torch.aten.add_.Tensor %1071, %arg52, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1073 = torch.aten.add.Tensor %1072, %result0_60, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1074 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_63, %result1_64, %result2_65 = torch.aten.native_layer_norm %1073, %1074, %arg51, %arg50, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1075 = torch.aten.t %arg61 : !torch.tensor -> !torch.tensor
    %1076 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1077 = torch.aten.view %result0_63, %1076 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1078 = torch.aten.mm %1077, %1075 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1079 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1080 = torch.aten._unsafe_view %1078, %1079 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1081 = torch.aten.add_.Tensor %1080, %arg60, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1082 = torch.aten.t %arg59 : !torch.tensor -> !torch.tensor
    %1083 = torch.aten.mm %1077, %1082 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1084 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1085 = torch.aten._unsafe_view %1083, %1084 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1086 = torch.aten.add_.Tensor %1085, %arg58, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1087 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1088 = torch.aten.view %1086, %1087 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1089 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1090 = torch.aten.permute %1088, %1089 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1091 = torch.aten.t %arg63 : !torch.tensor -> !torch.tensor
    %1092 = torch.aten.mm %1077, %1091 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1093 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1094 = torch.aten._unsafe_view %1092, %1093 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1095 = torch.aten.add_.Tensor %1094, %arg62, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1096 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1097 = torch.aten.view %1095, %1096 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1098 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1099 = torch.aten.permute %1097, %1098 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1100 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1101 = torch.aten.view %1081, %1100 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1102 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1103 = torch.aten.permute %1101, %1102 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1104 = torch.aten.transpose.int %1090, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1105 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1106 = torch.aten.expand %1103, %1105, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1107 = torch.aten.clone %1106, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1108 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1109 = torch.aten._unsafe_view %1107, %1108 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1110 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1111 = torch.aten.expand %1104, %1110, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1112 = torch.aten.clone %1111, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1113 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1114 = torch.aten._unsafe_view %1112, %1113 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1115 = torch.aten.bmm %1109, %1114 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1116 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1117 = torch.aten._unsafe_view %1115, %1116 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1118 = torch.aten.div.Tensor %1117, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1119 = torch.aten.add.Tensor %1118, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %1120 = torch.aten._softmax %1119, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %1121 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1122 = torch.aten.expand %1120, %1121, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1123 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1124 = torch.aten.view %1122, %1123 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1125 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1126 = torch.aten.expand %1099, %1125, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1127 = torch.aten.clone %1126, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1128 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1129 = torch.aten._unsafe_view %1127, %1128 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1130 = torch.aten.bmm %1124, %1129 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1131 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1132 = torch.aten._unsafe_view %1130, %1131 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1133 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1134 = torch.aten.permute %1132, %1133 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1135 = torch.aten.clone %1134, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1136 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1137 = torch.aten.view %1135, %1136 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1138 = torch.aten.t %arg57 : !torch.tensor -> !torch.tensor
    %1139 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1140 = torch.aten.view %1137, %1139 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1141 = torch.aten.mm %1140, %1138 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1142 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1143 = torch.aten._unsafe_view %1141, %1142 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1144 = torch.aten.add_.Tensor %1143, %arg56, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1145 = torch.aten.add.Tensor %1144, %result0_63, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1146 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_66, %result1_67, %result2_68 = torch.aten.native_layer_norm %1145, %1146, %arg55, %arg54, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1147 = torch.aten.t %arg65 : !torch.tensor -> !torch.tensor
    %1148 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1149 = torch.aten.view %result0_66, %1148 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1150 = torch.aten.mm %1149, %1147 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1151 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1152 = torch.aten._unsafe_view %1150, %1151 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1153 = torch.aten.add_.Tensor %1152, %arg64, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1154 = torch.aten.mul.Tensor %1153, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1155 = torch.aten.pow.Tensor_Scalar %1153, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %1156 = torch.aten.mul.Tensor %1155, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1157 = torch.aten.add.Tensor %1153, %1156, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1158 = torch.aten.mul.Tensor %1157, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1159 = torch.aten.tanh %1158 : !torch.tensor -> !torch.tensor
    %1160 = torch.aten.add.Tensor %1159, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %1161 = torch.aten.mul.Tensor %1154, %1160 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1162 = torch.aten.t %arg69 : !torch.tensor -> !torch.tensor
    %1163 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1164 = torch.aten.view %1161, %1163 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1165 = torch.aten.mm %1164, %1162 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1166 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1167 = torch.aten._unsafe_view %1165, %1166 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1168 = torch.aten.add_.Tensor %1167, %arg68, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1169 = torch.aten.add.Tensor %1168, %result0_66, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1170 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_69, %result1_70, %result2_71 = torch.aten.native_layer_norm %1169, %1170, %arg67, %arg66, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1171 = torch.aten.t %arg77 : !torch.tensor -> !torch.tensor
    %1172 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1173 = torch.aten.view %result0_69, %1172 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1174 = torch.aten.mm %1173, %1171 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1175 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1176 = torch.aten._unsafe_view %1174, %1175 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1177 = torch.aten.add_.Tensor %1176, %arg76, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1178 = torch.aten.t %arg75 : !torch.tensor -> !torch.tensor
    %1179 = torch.aten.mm %1173, %1178 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1180 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1181 = torch.aten._unsafe_view %1179, %1180 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1182 = torch.aten.add_.Tensor %1181, %arg74, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1183 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1184 = torch.aten.view %1182, %1183 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1185 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1186 = torch.aten.permute %1184, %1185 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1187 = torch.aten.t %arg79 : !torch.tensor -> !torch.tensor
    %1188 = torch.aten.mm %1173, %1187 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1189 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1190 = torch.aten._unsafe_view %1188, %1189 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1191 = torch.aten.add_.Tensor %1190, %arg78, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1192 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1193 = torch.aten.view %1191, %1192 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1194 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1195 = torch.aten.permute %1193, %1194 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1196 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1197 = torch.aten.view %1177, %1196 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1198 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1199 = torch.aten.permute %1197, %1198 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1200 = torch.aten.transpose.int %1186, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1201 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1202 = torch.aten.expand %1199, %1201, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1203 = torch.aten.clone %1202, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1204 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1205 = torch.aten._unsafe_view %1203, %1204 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1206 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1207 = torch.aten.expand %1200, %1206, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1208 = torch.aten.clone %1207, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1209 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1210 = torch.aten._unsafe_view %1208, %1209 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1211 = torch.aten.bmm %1205, %1210 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1212 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1213 = torch.aten._unsafe_view %1211, %1212 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1214 = torch.aten.div.Tensor %1213, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1215 = torch.aten.add.Tensor %1214, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %1216 = torch.aten._softmax %1215, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %1217 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1218 = torch.aten.expand %1216, %1217, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1219 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1220 = torch.aten.view %1218, %1219 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1221 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1222 = torch.aten.expand %1195, %1221, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1223 = torch.aten.clone %1222, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1224 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1225 = torch.aten._unsafe_view %1223, %1224 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1226 = torch.aten.bmm %1220, %1225 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1227 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1228 = torch.aten._unsafe_view %1226, %1227 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1229 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1230 = torch.aten.permute %1228, %1229 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1231 = torch.aten.clone %1230, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1232 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1233 = torch.aten.view %1231, %1232 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1234 = torch.aten.t %arg73 : !torch.tensor -> !torch.tensor
    %1235 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1236 = torch.aten.view %1233, %1235 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1237 = torch.aten.mm %1236, %1234 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1238 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1239 = torch.aten._unsafe_view %1237, %1238 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1240 = torch.aten.add_.Tensor %1239, %arg72, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1241 = torch.aten.add.Tensor %1240, %result0_69, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1242 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_72, %result1_73, %result2_74 = torch.aten.native_layer_norm %1241, %1242, %arg71, %arg70, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1243 = torch.aten.t %arg81 : !torch.tensor -> !torch.tensor
    %1244 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1245 = torch.aten.view %result0_72, %1244 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1246 = torch.aten.mm %1245, %1243 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1247 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1248 = torch.aten._unsafe_view %1246, %1247 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1249 = torch.aten.add_.Tensor %1248, %arg80, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1250 = torch.aten.mul.Tensor %1249, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1251 = torch.aten.pow.Tensor_Scalar %1249, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %1252 = torch.aten.mul.Tensor %1251, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1253 = torch.aten.add.Tensor %1249, %1252, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1254 = torch.aten.mul.Tensor %1253, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1255 = torch.aten.tanh %1254 : !torch.tensor -> !torch.tensor
    %1256 = torch.aten.add.Tensor %1255, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %1257 = torch.aten.mul.Tensor %1250, %1256 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1258 = torch.aten.t %arg85 : !torch.tensor -> !torch.tensor
    %1259 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1260 = torch.aten.view %1257, %1259 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1261 = torch.aten.mm %1260, %1258 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1262 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1263 = torch.aten._unsafe_view %1261, %1262 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1264 = torch.aten.add_.Tensor %1263, %arg84, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1265 = torch.aten.add.Tensor %1264, %result0_72, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1266 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_75, %result1_76, %result2_77 = torch.aten.native_layer_norm %1265, %1266, %arg83, %arg82, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1267 = torch.aten.t %arg93 : !torch.tensor -> !torch.tensor
    %1268 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1269 = torch.aten.view %result0_75, %1268 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1270 = torch.aten.mm %1269, %1267 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1271 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1272 = torch.aten._unsafe_view %1270, %1271 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1273 = torch.aten.add_.Tensor %1272, %arg92, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1274 = torch.aten.t %arg91 : !torch.tensor -> !torch.tensor
    %1275 = torch.aten.mm %1269, %1274 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1276 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1277 = torch.aten._unsafe_view %1275, %1276 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1278 = torch.aten.add_.Tensor %1277, %arg90, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1279 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1280 = torch.aten.view %1278, %1279 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1281 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1282 = torch.aten.permute %1280, %1281 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1283 = torch.aten.t %arg95 : !torch.tensor -> !torch.tensor
    %1284 = torch.aten.mm %1269, %1283 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1285 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1286 = torch.aten._unsafe_view %1284, %1285 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1287 = torch.aten.add_.Tensor %1286, %arg94, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1288 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1289 = torch.aten.view %1287, %1288 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1290 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1291 = torch.aten.permute %1289, %1290 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1292 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1293 = torch.aten.view %1273, %1292 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1294 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1295 = torch.aten.permute %1293, %1294 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1296 = torch.aten.transpose.int %1282, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1297 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1298 = torch.aten.expand %1295, %1297, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1299 = torch.aten.clone %1298, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1300 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1301 = torch.aten._unsafe_view %1299, %1300 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1302 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1303 = torch.aten.expand %1296, %1302, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1304 = torch.aten.clone %1303, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1305 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1306 = torch.aten._unsafe_view %1304, %1305 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1307 = torch.aten.bmm %1301, %1306 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1308 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1309 = torch.aten._unsafe_view %1307, %1308 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1310 = torch.aten.div.Tensor %1309, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1311 = torch.aten.add.Tensor %1310, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %1312 = torch.aten._softmax %1311, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %1313 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1314 = torch.aten.expand %1312, %1313, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1315 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1316 = torch.aten.view %1314, %1315 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1317 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1318 = torch.aten.expand %1291, %1317, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1319 = torch.aten.clone %1318, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1320 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1321 = torch.aten._unsafe_view %1319, %1320 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1322 = torch.aten.bmm %1316, %1321 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1323 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1324 = torch.aten._unsafe_view %1322, %1323 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1325 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1326 = torch.aten.permute %1324, %1325 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1327 = torch.aten.clone %1326, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1328 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1329 = torch.aten.view %1327, %1328 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1330 = torch.aten.t %arg89 : !torch.tensor -> !torch.tensor
    %1331 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1332 = torch.aten.view %1329, %1331 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1333 = torch.aten.mm %1332, %1330 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1334 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1335 = torch.aten._unsafe_view %1333, %1334 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1336 = torch.aten.add_.Tensor %1335, %arg88, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1337 = torch.aten.add.Tensor %1336, %result0_75, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1338 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_78, %result1_79, %result2_80 = torch.aten.native_layer_norm %1337, %1338, %arg87, %arg86, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1339 = torch.aten.t %arg97 : !torch.tensor -> !torch.tensor
    %1340 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1341 = torch.aten.view %result0_78, %1340 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1342 = torch.aten.mm %1341, %1339 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1343 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1344 = torch.aten._unsafe_view %1342, %1343 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1345 = torch.aten.add_.Tensor %1344, %arg96, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1346 = torch.aten.mul.Tensor %1345, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1347 = torch.aten.pow.Tensor_Scalar %1345, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %1348 = torch.aten.mul.Tensor %1347, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1349 = torch.aten.add.Tensor %1345, %1348, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1350 = torch.aten.mul.Tensor %1349, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1351 = torch.aten.tanh %1350 : !torch.tensor -> !torch.tensor
    %1352 = torch.aten.add.Tensor %1351, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %1353 = torch.aten.mul.Tensor %1346, %1352 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1354 = torch.aten.t %arg101 : !torch.tensor -> !torch.tensor
    %1355 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1356 = torch.aten.view %1353, %1355 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1357 = torch.aten.mm %1356, %1354 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1358 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1359 = torch.aten._unsafe_view %1357, %1358 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1360 = torch.aten.add_.Tensor %1359, %arg100, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1361 = torch.aten.add.Tensor %1360, %result0_78, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1362 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_81, %result1_82, %result2_83 = torch.aten.native_layer_norm %1361, %1362, %arg99, %arg98, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1363 = torch.aten.t %arg109 : !torch.tensor -> !torch.tensor
    %1364 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1365 = torch.aten.view %result0_81, %1364 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1366 = torch.aten.mm %1365, %1363 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1367 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1368 = torch.aten._unsafe_view %1366, %1367 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1369 = torch.aten.add_.Tensor %1368, %arg108, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1370 = torch.aten.t %arg107 : !torch.tensor -> !torch.tensor
    %1371 = torch.aten.mm %1365, %1370 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1372 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1373 = torch.aten._unsafe_view %1371, %1372 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1374 = torch.aten.add_.Tensor %1373, %arg106, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1375 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1376 = torch.aten.view %1374, %1375 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1377 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1378 = torch.aten.permute %1376, %1377 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1379 = torch.aten.t %arg111 : !torch.tensor -> !torch.tensor
    %1380 = torch.aten.mm %1365, %1379 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1381 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1382 = torch.aten._unsafe_view %1380, %1381 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1383 = torch.aten.add_.Tensor %1382, %arg110, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1384 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1385 = torch.aten.view %1383, %1384 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1386 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1387 = torch.aten.permute %1385, %1386 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1388 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1389 = torch.aten.view %1369, %1388 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1390 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1391 = torch.aten.permute %1389, %1390 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1392 = torch.aten.transpose.int %1378, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1393 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1394 = torch.aten.expand %1391, %1393, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1395 = torch.aten.clone %1394, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1396 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1397 = torch.aten._unsafe_view %1395, %1396 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1398 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1399 = torch.aten.expand %1392, %1398, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1400 = torch.aten.clone %1399, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1401 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1402 = torch.aten._unsafe_view %1400, %1401 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1403 = torch.aten.bmm %1397, %1402 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1404 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1405 = torch.aten._unsafe_view %1403, %1404 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1406 = torch.aten.div.Tensor %1405, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1407 = torch.aten.add.Tensor %1406, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %1408 = torch.aten._softmax %1407, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %1409 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1410 = torch.aten.expand %1408, %1409, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1411 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1412 = torch.aten.view %1410, %1411 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1413 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1414 = torch.aten.expand %1387, %1413, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1415 = torch.aten.clone %1414, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1416 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1417 = torch.aten._unsafe_view %1415, %1416 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1418 = torch.aten.bmm %1412, %1417 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1419 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1420 = torch.aten._unsafe_view %1418, %1419 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1421 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1422 = torch.aten.permute %1420, %1421 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1423 = torch.aten.clone %1422, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1424 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1425 = torch.aten.view %1423, %1424 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1426 = torch.aten.t %arg105 : !torch.tensor -> !torch.tensor
    %1427 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1428 = torch.aten.view %1425, %1427 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1429 = torch.aten.mm %1428, %1426 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1430 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1431 = torch.aten._unsafe_view %1429, %1430 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1432 = torch.aten.add_.Tensor %1431, %arg104, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1433 = torch.aten.add.Tensor %1432, %result0_81, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1434 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_84, %result1_85, %result2_86 = torch.aten.native_layer_norm %1433, %1434, %arg103, %arg102, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1435 = torch.aten.t %arg113 : !torch.tensor -> !torch.tensor
    %1436 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1437 = torch.aten.view %result0_84, %1436 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1438 = torch.aten.mm %1437, %1435 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1439 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1440 = torch.aten._unsafe_view %1438, %1439 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1441 = torch.aten.add_.Tensor %1440, %arg112, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1442 = torch.aten.mul.Tensor %1441, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1443 = torch.aten.pow.Tensor_Scalar %1441, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %1444 = torch.aten.mul.Tensor %1443, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1445 = torch.aten.add.Tensor %1441, %1444, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1446 = torch.aten.mul.Tensor %1445, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1447 = torch.aten.tanh %1446 : !torch.tensor -> !torch.tensor
    %1448 = torch.aten.add.Tensor %1447, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %1449 = torch.aten.mul.Tensor %1442, %1448 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1450 = torch.aten.t %arg117 : !torch.tensor -> !torch.tensor
    %1451 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1452 = torch.aten.view %1449, %1451 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1453 = torch.aten.mm %1452, %1450 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1454 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1455 = torch.aten._unsafe_view %1453, %1454 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1456 = torch.aten.add_.Tensor %1455, %arg116, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1457 = torch.aten.add.Tensor %1456, %result0_84, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1458 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_87, %result1_88, %result2_89 = torch.aten.native_layer_norm %1457, %1458, %arg115, %arg114, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1459 = torch.aten.t %arg125 : !torch.tensor -> !torch.tensor
    %1460 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1461 = torch.aten.view %result0_87, %1460 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1462 = torch.aten.mm %1461, %1459 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1463 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1464 = torch.aten._unsafe_view %1462, %1463 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1465 = torch.aten.add_.Tensor %1464, %arg124, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1466 = torch.aten.t %arg123 : !torch.tensor -> !torch.tensor
    %1467 = torch.aten.mm %1461, %1466 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1468 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1469 = torch.aten._unsafe_view %1467, %1468 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1470 = torch.aten.add_.Tensor %1469, %arg122, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1471 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1472 = torch.aten.view %1470, %1471 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1473 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1474 = torch.aten.permute %1472, %1473 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1475 = torch.aten.t %arg127 : !torch.tensor -> !torch.tensor
    %1476 = torch.aten.mm %1461, %1475 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1477 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1478 = torch.aten._unsafe_view %1476, %1477 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1479 = torch.aten.add_.Tensor %1478, %arg126, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1480 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1481 = torch.aten.view %1479, %1480 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1482 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1483 = torch.aten.permute %1481, %1482 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1484 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1485 = torch.aten.view %1465, %1484 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1486 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1487 = torch.aten.permute %1485, %1486 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1488 = torch.aten.transpose.int %1474, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1489 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1490 = torch.aten.expand %1487, %1489, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1491 = torch.aten.clone %1490, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1492 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1493 = torch.aten._unsafe_view %1491, %1492 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1494 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1495 = torch.aten.expand %1488, %1494, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1496 = torch.aten.clone %1495, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1497 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1498 = torch.aten._unsafe_view %1496, %1497 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1499 = torch.aten.bmm %1493, %1498 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1500 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1501 = torch.aten._unsafe_view %1499, %1500 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1502 = torch.aten.div.Tensor %1501, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1503 = torch.aten.add.Tensor %1502, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %1504 = torch.aten._softmax %1503, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %1505 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1506 = torch.aten.expand %1504, %1505, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1507 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1508 = torch.aten.view %1506, %1507 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1509 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1510 = torch.aten.expand %1483, %1509, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1511 = torch.aten.clone %1510, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1512 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1513 = torch.aten._unsafe_view %1511, %1512 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1514 = torch.aten.bmm %1508, %1513 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1515 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1516 = torch.aten._unsafe_view %1514, %1515 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1517 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1518 = torch.aten.permute %1516, %1517 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1519 = torch.aten.clone %1518, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1520 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1521 = torch.aten.view %1519, %1520 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1522 = torch.aten.t %arg121 : !torch.tensor -> !torch.tensor
    %1523 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1524 = torch.aten.view %1521, %1523 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1525 = torch.aten.mm %1524, %1522 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1526 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1527 = torch.aten._unsafe_view %1525, %1526 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1528 = torch.aten.add_.Tensor %1527, %arg120, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1529 = torch.aten.add.Tensor %1528, %result0_87, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1530 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_90, %result1_91, %result2_92 = torch.aten.native_layer_norm %1529, %1530, %arg119, %arg118, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1531 = torch.aten.t %arg129 : !torch.tensor -> !torch.tensor
    %1532 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1533 = torch.aten.view %result0_90, %1532 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1534 = torch.aten.mm %1533, %1531 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1535 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1536 = torch.aten._unsafe_view %1534, %1535 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1537 = torch.aten.add_.Tensor %1536, %arg128, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1538 = torch.aten.mul.Tensor %1537, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1539 = torch.aten.pow.Tensor_Scalar %1537, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %1540 = torch.aten.mul.Tensor %1539, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1541 = torch.aten.add.Tensor %1537, %1540, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1542 = torch.aten.mul.Tensor %1541, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1543 = torch.aten.tanh %1542 : !torch.tensor -> !torch.tensor
    %1544 = torch.aten.add.Tensor %1543, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %1545 = torch.aten.mul.Tensor %1538, %1544 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1546 = torch.aten.t %arg133 : !torch.tensor -> !torch.tensor
    %1547 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1548 = torch.aten.view %1545, %1547 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1549 = torch.aten.mm %1548, %1546 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1550 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1551 = torch.aten._unsafe_view %1549, %1550 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1552 = torch.aten.add_.Tensor %1551, %arg132, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1553 = torch.aten.add.Tensor %1552, %result0_90, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1554 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_93, %result1_94, %result2_95 = torch.aten.native_layer_norm %1553, %1554, %arg131, %arg130, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1555 = torch.aten.t %arg141 : !torch.tensor -> !torch.tensor
    %1556 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1557 = torch.aten.view %result0_93, %1556 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1558 = torch.aten.mm %1557, %1555 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1559 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1560 = torch.aten._unsafe_view %1558, %1559 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1561 = torch.aten.add_.Tensor %1560, %arg140, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1562 = torch.aten.t %arg139 : !torch.tensor -> !torch.tensor
    %1563 = torch.aten.mm %1557, %1562 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1564 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1565 = torch.aten._unsafe_view %1563, %1564 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1566 = torch.aten.add_.Tensor %1565, %arg138, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1567 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1568 = torch.aten.view %1566, %1567 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1569 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1570 = torch.aten.permute %1568, %1569 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1571 = torch.aten.t %arg143 : !torch.tensor -> !torch.tensor
    %1572 = torch.aten.mm %1557, %1571 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1573 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1574 = torch.aten._unsafe_view %1572, %1573 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1575 = torch.aten.add_.Tensor %1574, %arg142, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1576 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1577 = torch.aten.view %1575, %1576 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1578 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1579 = torch.aten.permute %1577, %1578 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1580 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1581 = torch.aten.view %1561, %1580 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1582 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1583 = torch.aten.permute %1581, %1582 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1584 = torch.aten.transpose.int %1570, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1585 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1586 = torch.aten.expand %1583, %1585, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1587 = torch.aten.clone %1586, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1588 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1589 = torch.aten._unsafe_view %1587, %1588 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1590 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1591 = torch.aten.expand %1584, %1590, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1592 = torch.aten.clone %1591, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1593 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1594 = torch.aten._unsafe_view %1592, %1593 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1595 = torch.aten.bmm %1589, %1594 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1596 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1597 = torch.aten._unsafe_view %1595, %1596 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1598 = torch.aten.div.Tensor %1597, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1599 = torch.aten.add.Tensor %1598, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %1600 = torch.aten._softmax %1599, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %1601 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1602 = torch.aten.expand %1600, %1601, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1603 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1604 = torch.aten.view %1602, %1603 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1605 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1606 = torch.aten.expand %1579, %1605, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1607 = torch.aten.clone %1606, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1608 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1609 = torch.aten._unsafe_view %1607, %1608 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1610 = torch.aten.bmm %1604, %1609 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1611 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1612 = torch.aten._unsafe_view %1610, %1611 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1613 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1614 = torch.aten.permute %1612, %1613 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1615 = torch.aten.clone %1614, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1616 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1617 = torch.aten.view %1615, %1616 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1618 = torch.aten.t %arg137 : !torch.tensor -> !torch.tensor
    %1619 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1620 = torch.aten.view %1617, %1619 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1621 = torch.aten.mm %1620, %1618 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1622 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1623 = torch.aten._unsafe_view %1621, %1622 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1624 = torch.aten.add_.Tensor %1623, %arg136, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1625 = torch.aten.add.Tensor %1624, %result0_93, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1626 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_96, %result1_97, %result2_98 = torch.aten.native_layer_norm %1625, %1626, %arg135, %arg134, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1627 = torch.aten.t %arg145 : !torch.tensor -> !torch.tensor
    %1628 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1629 = torch.aten.view %result0_96, %1628 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1630 = torch.aten.mm %1629, %1627 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1631 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1632 = torch.aten._unsafe_view %1630, %1631 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1633 = torch.aten.add_.Tensor %1632, %arg144, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1634 = torch.aten.mul.Tensor %1633, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1635 = torch.aten.pow.Tensor_Scalar %1633, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %1636 = torch.aten.mul.Tensor %1635, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1637 = torch.aten.add.Tensor %1633, %1636, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1638 = torch.aten.mul.Tensor %1637, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1639 = torch.aten.tanh %1638 : !torch.tensor -> !torch.tensor
    %1640 = torch.aten.add.Tensor %1639, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %1641 = torch.aten.mul.Tensor %1634, %1640 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1642 = torch.aten.t %arg149 : !torch.tensor -> !torch.tensor
    %1643 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1644 = torch.aten.view %1641, %1643 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1645 = torch.aten.mm %1644, %1642 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1646 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1647 = torch.aten._unsafe_view %1645, %1646 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1648 = torch.aten.add_.Tensor %1647, %arg148, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1649 = torch.aten.add.Tensor %1648, %result0_96, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1650 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_99, %result1_100, %result2_101 = torch.aten.native_layer_norm %1649, %1650, %arg147, %arg146, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1651 = torch.aten.t %arg157 : !torch.tensor -> !torch.tensor
    %1652 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1653 = torch.aten.view %result0_99, %1652 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1654 = torch.aten.mm %1653, %1651 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1655 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1656 = torch.aten._unsafe_view %1654, %1655 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1657 = torch.aten.add_.Tensor %1656, %arg156, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1658 = torch.aten.t %arg155 : !torch.tensor -> !torch.tensor
    %1659 = torch.aten.mm %1653, %1658 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1660 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1661 = torch.aten._unsafe_view %1659, %1660 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1662 = torch.aten.add_.Tensor %1661, %arg154, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1663 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1664 = torch.aten.view %1662, %1663 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1665 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1666 = torch.aten.permute %1664, %1665 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1667 = torch.aten.t %arg159 : !torch.tensor -> !torch.tensor
    %1668 = torch.aten.mm %1653, %1667 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1669 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1670 = torch.aten._unsafe_view %1668, %1669 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1671 = torch.aten.add_.Tensor %1670, %arg158, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1672 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1673 = torch.aten.view %1671, %1672 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1674 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1675 = torch.aten.permute %1673, %1674 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1676 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1677 = torch.aten.view %1657, %1676 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1678 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1679 = torch.aten.permute %1677, %1678 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1680 = torch.aten.transpose.int %1666, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1681 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1682 = torch.aten.expand %1679, %1681, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1683 = torch.aten.clone %1682, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1684 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1685 = torch.aten._unsafe_view %1683, %1684 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1686 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1687 = torch.aten.expand %1680, %1686, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1688 = torch.aten.clone %1687, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1689 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1690 = torch.aten._unsafe_view %1688, %1689 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1691 = torch.aten.bmm %1685, %1690 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1692 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1693 = torch.aten._unsafe_view %1691, %1692 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1694 = torch.aten.div.Tensor %1693, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1695 = torch.aten.add.Tensor %1694, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %1696 = torch.aten._softmax %1695, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %1697 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1698 = torch.aten.expand %1696, %1697, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1699 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1700 = torch.aten.view %1698, %1699 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1701 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1702 = torch.aten.expand %1675, %1701, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1703 = torch.aten.clone %1702, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1704 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1705 = torch.aten._unsafe_view %1703, %1704 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1706 = torch.aten.bmm %1700, %1705 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1707 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1708 = torch.aten._unsafe_view %1706, %1707 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1709 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1710 = torch.aten.permute %1708, %1709 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1711 = torch.aten.clone %1710, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1712 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1713 = torch.aten.view %1711, %1712 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1714 = torch.aten.t %arg153 : !torch.tensor -> !torch.tensor
    %1715 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1716 = torch.aten.view %1713, %1715 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1717 = torch.aten.mm %1716, %1714 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1718 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1719 = torch.aten._unsafe_view %1717, %1718 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1720 = torch.aten.add_.Tensor %1719, %arg152, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1721 = torch.aten.add.Tensor %1720, %result0_99, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1722 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_102, %result1_103, %result2_104 = torch.aten.native_layer_norm %1721, %1722, %arg151, %arg150, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1723 = torch.aten.t %arg161 : !torch.tensor -> !torch.tensor
    %1724 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1725 = torch.aten.view %result0_102, %1724 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1726 = torch.aten.mm %1725, %1723 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1727 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1728 = torch.aten._unsafe_view %1726, %1727 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1729 = torch.aten.add_.Tensor %1728, %arg160, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1730 = torch.aten.mul.Tensor %1729, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1731 = torch.aten.pow.Tensor_Scalar %1729, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %1732 = torch.aten.mul.Tensor %1731, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1733 = torch.aten.add.Tensor %1729, %1732, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1734 = torch.aten.mul.Tensor %1733, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1735 = torch.aten.tanh %1734 : !torch.tensor -> !torch.tensor
    %1736 = torch.aten.add.Tensor %1735, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %1737 = torch.aten.mul.Tensor %1730, %1736 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1738 = torch.aten.t %arg165 : !torch.tensor -> !torch.tensor
    %1739 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1740 = torch.aten.view %1737, %1739 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1741 = torch.aten.mm %1740, %1738 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1742 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1743 = torch.aten._unsafe_view %1741, %1742 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1744 = torch.aten.add_.Tensor %1743, %arg164, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1745 = torch.aten.add.Tensor %1744, %result0_102, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1746 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_105, %result1_106, %result2_107 = torch.aten.native_layer_norm %1745, %1746, %arg163, %arg162, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1747 = torch.aten.t %arg173 : !torch.tensor -> !torch.tensor
    %1748 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1749 = torch.aten.view %result0_105, %1748 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1750 = torch.aten.mm %1749, %1747 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1751 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1752 = torch.aten._unsafe_view %1750, %1751 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1753 = torch.aten.add_.Tensor %1752, %arg172, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1754 = torch.aten.t %arg171 : !torch.tensor -> !torch.tensor
    %1755 = torch.aten.mm %1749, %1754 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1756 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1757 = torch.aten._unsafe_view %1755, %1756 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1758 = torch.aten.add_.Tensor %1757, %arg170, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1759 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1760 = torch.aten.view %1758, %1759 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1761 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1762 = torch.aten.permute %1760, %1761 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1763 = torch.aten.t %arg175 : !torch.tensor -> !torch.tensor
    %1764 = torch.aten.mm %1749, %1763 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1765 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1766 = torch.aten._unsafe_view %1764, %1765 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1767 = torch.aten.add_.Tensor %1766, %arg174, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1768 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1769 = torch.aten.view %1767, %1768 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1770 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1771 = torch.aten.permute %1769, %1770 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1772 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1773 = torch.aten.view %1753, %1772 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1774 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1775 = torch.aten.permute %1773, %1774 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1776 = torch.aten.transpose.int %1762, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1777 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1778 = torch.aten.expand %1775, %1777, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1779 = torch.aten.clone %1778, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1780 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1781 = torch.aten._unsafe_view %1779, %1780 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1782 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1783 = torch.aten.expand %1776, %1782, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1784 = torch.aten.clone %1783, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1785 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1786 = torch.aten._unsafe_view %1784, %1785 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1787 = torch.aten.bmm %1781, %1786 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1788 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1789 = torch.aten._unsafe_view %1787, %1788 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1790 = torch.aten.div.Tensor %1789, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1791 = torch.aten.add.Tensor %1790, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %1792 = torch.aten._softmax %1791, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %1793 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1794 = torch.aten.expand %1792, %1793, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1795 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1796 = torch.aten.view %1794, %1795 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1797 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1798 = torch.aten.expand %1771, %1797, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1799 = torch.aten.clone %1798, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1800 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1801 = torch.aten._unsafe_view %1799, %1800 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1802 = torch.aten.bmm %1796, %1801 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1803 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1804 = torch.aten._unsafe_view %1802, %1803 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1805 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1806 = torch.aten.permute %1804, %1805 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1807 = torch.aten.clone %1806, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1808 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1809 = torch.aten.view %1807, %1808 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1810 = torch.aten.t %arg169 : !torch.tensor -> !torch.tensor
    %1811 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1812 = torch.aten.view %1809, %1811 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1813 = torch.aten.mm %1812, %1810 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1814 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1815 = torch.aten._unsafe_view %1813, %1814 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1816 = torch.aten.add_.Tensor %1815, %arg168, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1817 = torch.aten.add.Tensor %1816, %result0_105, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1818 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_108, %result1_109, %result2_110 = torch.aten.native_layer_norm %1817, %1818, %arg167, %arg166, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1819 = torch.aten.t %arg177 : !torch.tensor -> !torch.tensor
    %1820 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1821 = torch.aten.view %result0_108, %1820 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1822 = torch.aten.mm %1821, %1819 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1823 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1824 = torch.aten._unsafe_view %1822, %1823 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1825 = torch.aten.add_.Tensor %1824, %arg176, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1826 = torch.aten.mul.Tensor %1825, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1827 = torch.aten.pow.Tensor_Scalar %1825, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %1828 = torch.aten.mul.Tensor %1827, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1829 = torch.aten.add.Tensor %1825, %1828, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1830 = torch.aten.mul.Tensor %1829, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1831 = torch.aten.tanh %1830 : !torch.tensor -> !torch.tensor
    %1832 = torch.aten.add.Tensor %1831, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %1833 = torch.aten.mul.Tensor %1826, %1832 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1834 = torch.aten.t %arg181 : !torch.tensor -> !torch.tensor
    %1835 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1836 = torch.aten.view %1833, %1835 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1837 = torch.aten.mm %1836, %1834 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1838 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1839 = torch.aten._unsafe_view %1837, %1838 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1840 = torch.aten.add_.Tensor %1839, %arg180, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1841 = torch.aten.add.Tensor %1840, %result0_108, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1842 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_111, %result1_112, %result2_113 = torch.aten.native_layer_norm %1841, %1842, %arg179, %arg178, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1843 = torch.aten.t %arg189 : !torch.tensor -> !torch.tensor
    %1844 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1845 = torch.aten.view %result0_111, %1844 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1846 = torch.aten.mm %1845, %1843 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1847 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1848 = torch.aten._unsafe_view %1846, %1847 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1849 = torch.aten.add_.Tensor %1848, %arg188, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1850 = torch.aten.t %arg187 : !torch.tensor -> !torch.tensor
    %1851 = torch.aten.mm %1845, %1850 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1852 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1853 = torch.aten._unsafe_view %1851, %1852 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1854 = torch.aten.add_.Tensor %1853, %arg186, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1855 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1856 = torch.aten.view %1854, %1855 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1857 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1858 = torch.aten.permute %1856, %1857 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1859 = torch.aten.t %arg191 : !torch.tensor -> !torch.tensor
    %1860 = torch.aten.mm %1845, %1859 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1861 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1862 = torch.aten._unsafe_view %1860, %1861 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1863 = torch.aten.add_.Tensor %1862, %arg190, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1864 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1865 = torch.aten.view %1863, %1864 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1866 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1867 = torch.aten.permute %1865, %1866 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1868 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1869 = torch.aten.view %1849, %1868 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1870 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1871 = torch.aten.permute %1869, %1870 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1872 = torch.aten.transpose.int %1858, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1873 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1874 = torch.aten.expand %1871, %1873, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1875 = torch.aten.clone %1874, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1876 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1877 = torch.aten._unsafe_view %1875, %1876 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1878 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1879 = torch.aten.expand %1872, %1878, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1880 = torch.aten.clone %1879, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1881 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1882 = torch.aten._unsafe_view %1880, %1881 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1883 = torch.aten.bmm %1877, %1882 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1884 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1885 = torch.aten._unsafe_view %1883, %1884 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1886 = torch.aten.div.Tensor %1885, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1887 = torch.aten.add.Tensor %1886, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %1888 = torch.aten._softmax %1887, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %1889 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1890 = torch.aten.expand %1888, %1889, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1891 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1892 = torch.aten.view %1890, %1891 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1893 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1894 = torch.aten.expand %1867, %1893, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1895 = torch.aten.clone %1894, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1896 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1897 = torch.aten._unsafe_view %1895, %1896 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1898 = torch.aten.bmm %1892, %1897 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1899 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1900 = torch.aten._unsafe_view %1898, %1899 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1901 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1902 = torch.aten.permute %1900, %1901 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1903 = torch.aten.clone %1902, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1904 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1905 = torch.aten.view %1903, %1904 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1906 = torch.aten.t %arg185 : !torch.tensor -> !torch.tensor
    %1907 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1908 = torch.aten.view %1905, %1907 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1909 = torch.aten.mm %1908, %1906 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1910 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1911 = torch.aten._unsafe_view %1909, %1910 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1912 = torch.aten.add_.Tensor %1911, %arg184, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1913 = torch.aten.add.Tensor %1912, %result0_111, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1914 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_114, %result1_115, %result2_116 = torch.aten.native_layer_norm %1913, %1914, %arg183, %arg182, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1915 = torch.aten.t %arg193 : !torch.tensor -> !torch.tensor
    %1916 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1917 = torch.aten.view %result0_114, %1916 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1918 = torch.aten.mm %1917, %1915 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1919 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1920 = torch.aten._unsafe_view %1918, %1919 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1921 = torch.aten.add_.Tensor %1920, %arg192, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1922 = torch.aten.mul.Tensor %1921, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1923 = torch.aten.pow.Tensor_Scalar %1921, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %1924 = torch.aten.mul.Tensor %1923, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1925 = torch.aten.add.Tensor %1921, %1924, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1926 = torch.aten.mul.Tensor %1925, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1927 = torch.aten.tanh %1926 : !torch.tensor -> !torch.tensor
    %1928 = torch.aten.add.Tensor %1927, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %1929 = torch.aten.mul.Tensor %1922, %1928 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1930 = torch.aten.t %arg197 : !torch.tensor -> !torch.tensor
    %1931 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1932 = torch.aten.view %1929, %1931 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1933 = torch.aten.mm %1932, %1930 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1934 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1935 = torch.aten._unsafe_view %1933, %1934 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1936 = torch.aten.add_.Tensor %1935, %arg196, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1937 = torch.aten.add.Tensor %1936, %result0_114, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1938 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_117, %result1_118, %result2_119 = torch.aten.native_layer_norm %1937, %1938, %arg195, %arg194, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %1939 = torch.aten.t %arg221 : !torch.tensor -> !torch.tensor
    %1940 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %1941 = torch.aten.view %result0_117, %1940 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1942 = torch.aten.mm %1941, %1939 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1943 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1944 = torch.aten._unsafe_view %1942, %1943 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1945 = torch.aten.add_.Tensor %1944, %arg220, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1946 = torch.aten.t %arg219 : !torch.tensor -> !torch.tensor
    %1947 = torch.aten.mm %1941, %1946 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1948 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1949 = torch.aten._unsafe_view %1947, %1948 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1950 = torch.aten.add_.Tensor %1949, %arg218, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1951 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1952 = torch.aten.view %1950, %1951 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1953 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1954 = torch.aten.permute %1952, %1953 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1955 = torch.aten.t %arg223 : !torch.tensor -> !torch.tensor
    %1956 = torch.aten.mm %1941, %1955 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1957 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1958 = torch.aten._unsafe_view %1956, %1957 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1959 = torch.aten.add_.Tensor %1958, %arg222, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %1960 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1961 = torch.aten.view %1959, %1960 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1962 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1963 = torch.aten.permute %1961, %1962 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1964 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1965 = torch.aten.view %1945, %1964 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1966 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1967 = torch.aten.permute %1965, %1966 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1968 = torch.aten.transpose.int %1954, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %1969 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1970 = torch.aten.expand %1967, %1969, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1971 = torch.aten.clone %1970, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1972 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1973 = torch.aten._unsafe_view %1971, %1972 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1974 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1975 = torch.aten.expand %1968, %1974, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1976 = torch.aten.clone %1975, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1977 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1978 = torch.aten._unsafe_view %1976, %1977 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1979 = torch.aten.bmm %1973, %1978 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1980 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1981 = torch.aten._unsafe_view %1979, %1980 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1982 = torch.aten.div.Tensor %1981, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %1983 = torch.aten.add.Tensor %1982, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %1984 = torch.aten._softmax %1983, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %1985 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1986 = torch.aten.expand %1984, %1985, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1987 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1988 = torch.aten.view %1986, %1987 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1989 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1990 = torch.aten.expand %1963, %1989, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %1991 = torch.aten.clone %1990, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %1992 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1993 = torch.aten._unsafe_view %1991, %1992 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1994 = torch.aten.bmm %1988, %1993 : !torch.tensor, !torch.tensor -> !torch.tensor
    %1995 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1996 = torch.aten._unsafe_view %1994, %1995 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1997 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %1998 = torch.aten.permute %1996, %1997 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %1999 = torch.aten.clone %1998, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %2000 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2001 = torch.aten.view %1999, %2000 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2002 = torch.aten.t %arg217 : !torch.tensor -> !torch.tensor
    %2003 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %2004 = torch.aten.view %2001, %2003 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2005 = torch.aten.mm %2004, %2002 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2006 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2007 = torch.aten._unsafe_view %2005, %2006 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2008 = torch.aten.add_.Tensor %2007, %arg216, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2009 = torch.aten.add.Tensor %2008, %result0_117, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2010 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_120, %result1_121, %result2_122 = torch.aten.native_layer_norm %2009, %2010, %arg215, %arg214, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %2011 = torch.aten.t %arg225 : !torch.tensor -> !torch.tensor
    %2012 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %2013 = torch.aten.view %result0_120, %2012 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2014 = torch.aten.mm %2013, %2011 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2015 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2016 = torch.aten._unsafe_view %2014, %2015 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2017 = torch.aten.add_.Tensor %2016, %arg224, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2018 = torch.aten.mul.Tensor %2017, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %2019 = torch.aten.pow.Tensor_Scalar %2017, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %2020 = torch.aten.mul.Tensor %2019, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %2021 = torch.aten.add.Tensor %2017, %2020, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2022 = torch.aten.mul.Tensor %2021, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %2023 = torch.aten.tanh %2022 : !torch.tensor -> !torch.tensor
    %2024 = torch.aten.add.Tensor %2023, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %2025 = torch.aten.mul.Tensor %2018, %2024 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2026 = torch.aten.t %arg229 : !torch.tensor -> !torch.tensor
    %2027 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %2028 = torch.aten.view %2025, %2027 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2029 = torch.aten.mm %2028, %2026 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2030 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2031 = torch.aten._unsafe_view %2029, %2030 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2032 = torch.aten.add_.Tensor %2031, %arg228, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2033 = torch.aten.add.Tensor %2032, %result0_120, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2034 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_123, %result1_124, %result2_125 = torch.aten.native_layer_norm %2033, %2034, %arg227, %arg226, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %2035 = torch.aten.t %arg237 : !torch.tensor -> !torch.tensor
    %2036 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %2037 = torch.aten.view %result0_123, %2036 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2038 = torch.aten.mm %2037, %2035 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2039 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2040 = torch.aten._unsafe_view %2038, %2039 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2041 = torch.aten.add_.Tensor %2040, %arg236, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2042 = torch.aten.t %arg235 : !torch.tensor -> !torch.tensor
    %2043 = torch.aten.mm %2037, %2042 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2044 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2045 = torch.aten._unsafe_view %2043, %2044 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2046 = torch.aten.add_.Tensor %2045, %arg234, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2047 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2048 = torch.aten.view %2046, %2047 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2049 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2050 = torch.aten.permute %2048, %2049 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2051 = torch.aten.t %arg239 : !torch.tensor -> !torch.tensor
    %2052 = torch.aten.mm %2037, %2051 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2053 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2054 = torch.aten._unsafe_view %2052, %2053 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2055 = torch.aten.add_.Tensor %2054, %arg238, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2056 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2057 = torch.aten.view %2055, %2056 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2058 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2059 = torch.aten.permute %2057, %2058 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2060 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2061 = torch.aten.view %2041, %2060 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2062 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2063 = torch.aten.permute %2061, %2062 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2064 = torch.aten.transpose.int %2050, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %2065 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2066 = torch.aten.expand %2063, %2065, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %2067 = torch.aten.clone %2066, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %2068 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2069 = torch.aten._unsafe_view %2067, %2068 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2070 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2071 = torch.aten.expand %2064, %2070, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %2072 = torch.aten.clone %2071, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %2073 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2074 = torch.aten._unsafe_view %2072, %2073 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2075 = torch.aten.bmm %2069, %2074 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2076 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2077 = torch.aten._unsafe_view %2075, %2076 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2078 = torch.aten.div.Tensor %2077, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %2079 = torch.aten.add.Tensor %2078, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %2080 = torch.aten._softmax %2079, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %2081 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2082 = torch.aten.expand %2080, %2081, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %2083 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2084 = torch.aten.view %2082, %2083 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2085 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2086 = torch.aten.expand %2059, %2085, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %2087 = torch.aten.clone %2086, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %2088 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2089 = torch.aten._unsafe_view %2087, %2088 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2090 = torch.aten.bmm %2084, %2089 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2091 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2092 = torch.aten._unsafe_view %2090, %2091 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2093 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2094 = torch.aten.permute %2092, %2093 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2095 = torch.aten.clone %2094, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %2096 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2097 = torch.aten.view %2095, %2096 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2098 = torch.aten.t %arg233 : !torch.tensor -> !torch.tensor
    %2099 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %2100 = torch.aten.view %2097, %2099 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2101 = torch.aten.mm %2100, %2098 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2102 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2103 = torch.aten._unsafe_view %2101, %2102 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2104 = torch.aten.add_.Tensor %2103, %arg232, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2105 = torch.aten.add.Tensor %2104, %result0_123, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2106 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_126, %result1_127, %result2_128 = torch.aten.native_layer_norm %2105, %2106, %arg231, %arg230, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %2107 = torch.aten.t %arg241 : !torch.tensor -> !torch.tensor
    %2108 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %2109 = torch.aten.view %result0_126, %2108 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2110 = torch.aten.mm %2109, %2107 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2111 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2112 = torch.aten._unsafe_view %2110, %2111 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2113 = torch.aten.add_.Tensor %2112, %arg240, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2114 = torch.aten.mul.Tensor %2113, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %2115 = torch.aten.pow.Tensor_Scalar %2113, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %2116 = torch.aten.mul.Tensor %2115, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %2117 = torch.aten.add.Tensor %2113, %2116, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2118 = torch.aten.mul.Tensor %2117, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %2119 = torch.aten.tanh %2118 : !torch.tensor -> !torch.tensor
    %2120 = torch.aten.add.Tensor %2119, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %2121 = torch.aten.mul.Tensor %2114, %2120 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2122 = torch.aten.t %arg245 : !torch.tensor -> !torch.tensor
    %2123 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %2124 = torch.aten.view %2121, %2123 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2125 = torch.aten.mm %2124, %2122 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2126 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2127 = torch.aten._unsafe_view %2125, %2126 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2128 = torch.aten.add_.Tensor %2127, %arg244, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2129 = torch.aten.add.Tensor %2128, %result0_126, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2130 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_129, %result1_130, %result2_131 = torch.aten.native_layer_norm %2129, %2130, %arg243, %arg242, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %2131 = torch.aten.t %arg253 : !torch.tensor -> !torch.tensor
    %2132 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %2133 = torch.aten.view %result0_129, %2132 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2134 = torch.aten.mm %2133, %2131 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2135 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2136 = torch.aten._unsafe_view %2134, %2135 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2137 = torch.aten.add_.Tensor %2136, %arg252, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2138 = torch.aten.t %arg251 : !torch.tensor -> !torch.tensor
    %2139 = torch.aten.mm %2133, %2138 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2140 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2141 = torch.aten._unsafe_view %2139, %2140 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2142 = torch.aten.add_.Tensor %2141, %arg250, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2143 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2144 = torch.aten.view %2142, %2143 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2145 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2146 = torch.aten.permute %2144, %2145 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2147 = torch.aten.t %arg255 : !torch.tensor -> !torch.tensor
    %2148 = torch.aten.mm %2133, %2147 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2149 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2150 = torch.aten._unsafe_view %2148, %2149 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2151 = torch.aten.add_.Tensor %2150, %arg254, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2152 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2153 = torch.aten.view %2151, %2152 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2154 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2155 = torch.aten.permute %2153, %2154 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2156 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2157 = torch.aten.view %2137, %2156 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2158 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2159 = torch.aten.permute %2157, %2158 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2160 = torch.aten.transpose.int %2146, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %2161 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2162 = torch.aten.expand %2159, %2161, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %2163 = torch.aten.clone %2162, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %2164 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2165 = torch.aten._unsafe_view %2163, %2164 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2166 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2167 = torch.aten.expand %2160, %2166, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %2168 = torch.aten.clone %2167, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %2169 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2170 = torch.aten._unsafe_view %2168, %2169 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2171 = torch.aten.bmm %2165, %2170 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2172 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2173 = torch.aten._unsafe_view %2171, %2172 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2174 = torch.aten.div.Tensor %2173, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %2175 = torch.aten.add.Tensor %2174, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %2176 = torch.aten._softmax %2175, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %2177 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2178 = torch.aten.expand %2176, %2177, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %2179 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2180 = torch.aten.view %2178, %2179 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2181 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2182 = torch.aten.expand %2155, %2181, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %2183 = torch.aten.clone %2182, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %2184 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2185 = torch.aten._unsafe_view %2183, %2184 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2186 = torch.aten.bmm %2180, %2185 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2187 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2188 = torch.aten._unsafe_view %2186, %2187 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2189 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2190 = torch.aten.permute %2188, %2189 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2191 = torch.aten.clone %2190, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %2192 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2193 = torch.aten.view %2191, %2192 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2194 = torch.aten.t %arg249 : !torch.tensor -> !torch.tensor
    %2195 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %2196 = torch.aten.view %2193, %2195 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2197 = torch.aten.mm %2196, %2194 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2198 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2199 = torch.aten._unsafe_view %2197, %2198 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2200 = torch.aten.add_.Tensor %2199, %arg248, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2201 = torch.aten.add.Tensor %2200, %result0_129, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2202 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_132, %result1_133, %result2_134 = torch.aten.native_layer_norm %2201, %2202, %arg247, %arg246, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %2203 = torch.aten.t %arg257 : !torch.tensor -> !torch.tensor
    %2204 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %2205 = torch.aten.view %result0_132, %2204 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2206 = torch.aten.mm %2205, %2203 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2207 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2208 = torch.aten._unsafe_view %2206, %2207 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2209 = torch.aten.add_.Tensor %2208, %arg256, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2210 = torch.aten.mul.Tensor %2209, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %2211 = torch.aten.pow.Tensor_Scalar %2209, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %2212 = torch.aten.mul.Tensor %2211, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %2213 = torch.aten.add.Tensor %2209, %2212, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2214 = torch.aten.mul.Tensor %2213, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %2215 = torch.aten.tanh %2214 : !torch.tensor -> !torch.tensor
    %2216 = torch.aten.add.Tensor %2215, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %2217 = torch.aten.mul.Tensor %2210, %2216 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2218 = torch.aten.t %arg261 : !torch.tensor -> !torch.tensor
    %2219 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %2220 = torch.aten.view %2217, %2219 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2221 = torch.aten.mm %2220, %2218 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2222 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2223 = torch.aten._unsafe_view %2221, %2222 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2224 = torch.aten.add_.Tensor %2223, %arg260, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2225 = torch.aten.add.Tensor %2224, %result0_132, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2226 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_135, %result1_136, %result2_137 = torch.aten.native_layer_norm %2225, %2226, %arg259, %arg258, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %2227 = torch.aten.t %arg269 : !torch.tensor -> !torch.tensor
    %2228 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %2229 = torch.aten.view %result0_135, %2228 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2230 = torch.aten.mm %2229, %2227 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2231 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2232 = torch.aten._unsafe_view %2230, %2231 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2233 = torch.aten.add_.Tensor %2232, %arg268, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2234 = torch.aten.t %arg267 : !torch.tensor -> !torch.tensor
    %2235 = torch.aten.mm %2229, %2234 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2236 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2237 = torch.aten._unsafe_view %2235, %2236 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2238 = torch.aten.add_.Tensor %2237, %arg266, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2239 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2240 = torch.aten.view %2238, %2239 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2241 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2242 = torch.aten.permute %2240, %2241 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2243 = torch.aten.t %arg271 : !torch.tensor -> !torch.tensor
    %2244 = torch.aten.mm %2229, %2243 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2245 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2246 = torch.aten._unsafe_view %2244, %2245 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2247 = torch.aten.add_.Tensor %2246, %arg270, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2248 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2249 = torch.aten.view %2247, %2248 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2250 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2251 = torch.aten.permute %2249, %2250 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2252 = torch.prim.ListConstruct %int4, %int512, %int16, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2253 = torch.aten.view %2233, %2252 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2254 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2255 = torch.aten.permute %2253, %2254 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2256 = torch.aten.transpose.int %2242, %int-1, %int-2 : !torch.tensor, !torch.int, !torch.int -> !torch.tensor
    %2257 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2258 = torch.aten.expand %2255, %2257, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %2259 = torch.aten.clone %2258, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %2260 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2261 = torch.aten._unsafe_view %2259, %2260 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2262 = torch.prim.ListConstruct %int4, %int16, %int64, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2263 = torch.aten.expand %2256, %2262, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %2264 = torch.aten.clone %2263, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %2265 = torch.prim.ListConstruct %int64, %int64, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2266 = torch.aten._unsafe_view %2264, %2265 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2267 = torch.aten.bmm %2261, %2266 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2268 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2269 = torch.aten._unsafe_view %2267, %2268 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2270 = torch.aten.div.Tensor %2269, %6 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %2271 = torch.aten.add.Tensor %2270, %5, %int1 : !torch.tensor, !torch.tensor<[4,1,1,512],f32>, !torch.int -> !torch.tensor
    %2272 = torch.aten._softmax %2271, %int-1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %2273 = torch.prim.ListConstruct %int4, %int16, %int512, %int512 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2274 = torch.aten.expand %2272, %2273, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %2275 = torch.prim.ListConstruct %int64, %int512, %int512 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2276 = torch.aten.view %2274, %2275 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2277 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2278 = torch.aten.expand %2251, %2277, %false : !torch.tensor, !torch.list<!torch.int>, !torch.bool -> !torch.tensor
    %2279 = torch.aten.clone %2278, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %2280 = torch.prim.ListConstruct %int64, %int512, %int64 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2281 = torch.aten._unsafe_view %2279, %2280 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2282 = torch.aten.bmm %2276, %2281 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2283 = torch.prim.ListConstruct %int4, %int16, %int512, %int64 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2284 = torch.aten._unsafe_view %2282, %2283 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2285 = torch.prim.ListConstruct %int0, %int2, %int1, %int3 : (!torch.int, !torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2286 = torch.aten.permute %2284, %2285 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2287 = torch.aten.clone %2286, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %2288 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2289 = torch.aten.view %2287, %2288 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2290 = torch.aten.t %arg265 : !torch.tensor -> !torch.tensor
    %2291 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %2292 = torch.aten.view %2289, %2291 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2293 = torch.aten.mm %2292, %2290 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2294 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2295 = torch.aten._unsafe_view %2293, %2294 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2296 = torch.aten.add_.Tensor %2295, %arg264, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2297 = torch.aten.add.Tensor %2296, %result0_135, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2298 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_138, %result1_139, %result2_140 = torch.aten.native_layer_norm %2297, %2298, %arg263, %arg262, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %2299 = torch.aten.t %arg273 : !torch.tensor -> !torch.tensor
    %2300 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %2301 = torch.aten.view %result0_138, %2300 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2302 = torch.aten.mm %2301, %2299 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2303 = torch.prim.ListConstruct %int4, %int512, %int4096 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2304 = torch.aten._unsafe_view %2302, %2303 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2305 = torch.aten.add_.Tensor %2304, %arg272, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2306 = torch.aten.mul.Tensor %2305, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %2307 = torch.aten.pow.Tensor_Scalar %2305, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %2308 = torch.aten.mul.Tensor %2307, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %2309 = torch.aten.add.Tensor %2305, %2308, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2310 = torch.aten.mul.Tensor %2309, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %2311 = torch.aten.tanh %2310 : !torch.tensor -> !torch.tensor
    %2312 = torch.aten.add.Tensor %2311, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %2313 = torch.aten.mul.Tensor %2306, %2312 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2314 = torch.aten.t %arg277 : !torch.tensor -> !torch.tensor
    %2315 = torch.prim.ListConstruct %int2048, %int4096 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %2316 = torch.aten.view %2313, %2315 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2317 = torch.aten.mm %2316, %2314 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2318 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2319 = torch.aten._unsafe_view %2317, %2318 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2320 = torch.aten.add_.Tensor %2319, %arg276, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2321 = torch.aten.add.Tensor %2320, %result0_138, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2322 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_141, %result1_142, %result2_143 = torch.aten.native_layer_norm %2321, %2322, %arg275, %arg274, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %2323 = torch.aten.t %arg398 : !torch.tensor -> !torch.tensor
    %2324 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %2325 = torch.aten.view %result0_141, %2324 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2326 = torch.aten.mm %2325, %2323 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2327 = torch.prim.ListConstruct %int4, %int512, %int1024 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2328 = torch.aten._unsafe_view %2326, %2327 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2329 = torch.aten.add_.Tensor %2328, %arg397, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2330 = torch.aten.mul.Tensor %2329, %4 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %2331 = torch.aten.pow.Tensor_Scalar %2329, %float3.000000e00 : !torch.tensor, !torch.float -> !torch.tensor
    %2332 = torch.aten.mul.Tensor %2331, %3 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %2333 = torch.aten.add.Tensor %2329, %2332, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2334 = torch.aten.mul.Tensor %2333, %2 : !torch.tensor, !torch.tensor<[],f64> -> !torch.tensor
    %2335 = torch.aten.tanh %2334 : !torch.tensor -> !torch.tensor
    %2336 = torch.aten.add.Tensor %2335, %1, %int1 : !torch.tensor, !torch.tensor<[],f64>, !torch.int -> !torch.tensor
    %2337 = torch.aten.mul.Tensor %2330, %2336 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2338 = torch.prim.ListConstruct %int1024 : (!torch.int) -> !torch.list<!torch.int>
    %result0_144, %result1_145, %result2_146 = torch.aten.native_layer_norm %2337, %2338, %arg396, %arg395, %float9.999990e-13 : !torch.tensor, !torch.list<!torch.int>, !torch.tensor, !torch.tensor, !torch.float -> !torch.tensor, !torch.tensor, !torch.tensor
    %2339 = torch.aten.t %arg394 : !torch.tensor -> !torch.tensor
    %2340 = torch.prim.ListConstruct %int2048, %int1024 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %2341 = torch.aten.view %result0_144, %2340 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2342 = torch.aten.mm %2341, %2339 : !torch.tensor, !torch.tensor -> !torch.tensor
    %2343 = torch.prim.ListConstruct %int4, %int512, %int50358 : (!torch.int, !torch.int, !torch.int) -> !torch.list<!torch.int>
    %2344 = torch.aten._unsafe_view %2342, %2343 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2345 = torch.aten.add_.Tensor %2344, %arg393, %int1 : !torch.tensor, !torch.tensor, !torch.int -> !torch.tensor
    %2346 = torch.aten.slice.Tensor %2345, %int0, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %2347 = torch.aten.slice.Tensor %2346, %int1, %int0, %int-1, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %2348 = torch.aten.slice.Tensor %2347, %int2, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %2349 = torch.aten.clone %2348, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %2350 = torch.aten.slice.Tensor %arg402, %int0, %int0, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %2351 = torch.aten.slice.Tensor %2350, %int1, %int1, %int9223372036854775807, %int1 : !torch.tensor, !torch.int, !torch.int, !torch.int, !torch.int -> !torch.tensor
    %2352 = torch.aten.clone %2351, %int0 : !torch.tensor, !torch.int -> !torch.tensor
    %2353 = torch.prim.ListConstruct %int-1, %int50358 : (!torch.int, !torch.int) -> !torch.list<!torch.int>
    %2354 = torch.aten.view %2349, %2353 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2355 = torch.prim.ListConstruct %int-1 : (!torch.int) -> !torch.list<!torch.int>
    %2356 = torch.aten.view %2352, %2355 : !torch.tensor, !torch.list<!torch.int> -> !torch.tensor
    %2357 = torch.aten._log_softmax %2354, %int1, %false : !torch.tensor, !torch.int, !torch.bool -> !torch.tensor
    %output, %total_weight = torch.aten.nll_loss_forward %2357, %2356, %none, %int1, %int-100 : !torch.tensor, !torch.tensor, !torch.none, !torch.int, !torch.int -> !torch.tensor, !torch.tensor
    %2358 = torch.prim.ListConstruct %output, %2345 : (!torch.tensor, !torch.tensor) -> !torch.list<!torch.tensor>
    return %2358 : !torch.list<!torch.tensor>
  }
  torch.class_type @__torch__.torch.fx.graph_module.___torch_mangle_0.GraphModule {
    torch.method "forward", @__torch__.torch.fx.graph_module.___torch_mangle_0.GraphModule.forward
  }
  %0 = torch.nn_module {
  } : !torch.nn.Module<"__torch__.torch.fx.graph_module.___torch_mangle_0.GraphModule">
}