File size: 197,123 Bytes
f24f82b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 |
program(1.0)
[buildInfo = dict<tensor<string, []>, tensor<string, []>>({{"coremlc-component-MIL", "3510.2.1"}, {"coremlc-version", "3500.32.1"}, {"coremltools-component-torch", "2.9.1"}, {"coremltools-source-dialect", "TorchScript"}, {"coremltools-version", "9.0"}})]
{
func main<ios15>(tensor<int32, [1, 77]> text) {
tensor<int32, []> var_17_axis_0 = const()[name = tensor<string, []>("op_17_axis_0"), val = tensor<int32, []>(0)];
tensor<fp16, [49408, 512]> model_token_embedding_weight_to_fp16 = const()[name = tensor<string, []>("model_token_embedding_weight_to_fp16"), val = tensor<fp16, [49408, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(64)))];
tensor<fp16, [1, 77, 512]> var_17_cast_fp16 = gather(axis = var_17_axis_0, indices = text, x = model_token_embedding_weight_to_fp16)[name = tensor<string, []>("op_17_cast_fp16")];
tensor<fp16, [77, 512]> const_0_to_fp16 = const()[name = tensor<string, []>("const_0_to_fp16"), val = tensor<fp16, [77, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(50593920)))];
tensor<fp16, [1, 77, 512]> input_1_cast_fp16 = add(x = var_17_cast_fp16, y = const_0_to_fp16)[name = tensor<string, []>("input_1_cast_fp16")];
tensor<int32, [1]> x_3_axes_0 = const()[name = tensor<string, []>("x_3_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_0_ln_1_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_0_ln_1_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(50672832)))];
tensor<fp16, [512]> model_transformer_resblocks_0_ln_1_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_0_ln_1_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(50673920)))];
tensor<fp16, []> var_34_to_fp16 = const()[name = tensor<string, []>("op_34_to_fp16"), val = tensor<fp16, []>(0x1.5p-17)];
tensor<fp16, [1, 77, 512]> x_3_cast_fp16 = layer_norm(axes = x_3_axes_0, beta = model_transformer_resblocks_0_ln_1_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_0_ln_1_weight_to_fp16, x = input_1_cast_fp16)[name = tensor<string, []>("x_3_cast_fp16")];
tensor<int32, [3]> query_3_perm_0 = const()[name = tensor<string, []>("query_3_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1536, 512]> model_transformer_resblocks_0_attn_in_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_0_attn_in_proj_weight_to_fp16"), val = tensor<fp16, [1536, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(50675008)))];
tensor<fp16, [1536]> model_transformer_resblocks_0_attn_in_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_0_attn_in_proj_bias_to_fp16"), val = tensor<fp16, [1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(52247936)))];
tensor<fp16, [77, 1, 512]> query_3_cast_fp16 = transpose(perm = query_3_perm_0, x = x_3_cast_fp16)[name = tensor<string, []>("transpose_85")];
tensor<fp16, [77, 1, 1536]> linear_0_cast_fp16 = linear(bias = model_transformer_resblocks_0_attn_in_proj_bias_to_fp16, weight = model_transformer_resblocks_0_attn_in_proj_weight_to_fp16, x = query_3_cast_fp16)[name = tensor<string, []>("linear_0_cast_fp16")];
tensor<int32, [4]> concat_0 = const()[name = tensor<string, []>("concat_0"), val = tensor<int32, [4]>([77, 1, 3, 512])];
tensor<fp16, [77, 1, 3, 512]> var_104_cast_fp16 = reshape(shape = concat_0, x = linear_0_cast_fp16)[name = tensor<string, []>("op_104_cast_fp16")];
tensor<int32, [1]> var_105_axes_0 = const()[name = tensor<string, []>("op_105_axes_0"), val = tensor<int32, [1]>([0])];
tensor<fp16, [1, 77, 1, 3, 512]> var_105_cast_fp16 = expand_dims(axes = var_105_axes_0, x = var_104_cast_fp16)[name = tensor<string, []>("op_105_cast_fp16")];
tensor<int32, [5]> var_106_perm_0 = const()[name = tensor<string, []>("op_106_perm_0"), val = tensor<int32, [5]>([-2, 1, 2, 0, 4])];
tensor<int32, [1]> var_107_axes_0 = const()[name = tensor<string, []>("op_107_axes_0"), val = tensor<int32, [1]>([-2])];
tensor<fp16, [3, 77, 1, 1, 512]> var_106_cast_fp16 = transpose(perm = var_106_perm_0, x = var_105_cast_fp16)[name = tensor<string, []>("transpose_84")];
tensor<fp16, [3, 77, 1, 512]> var_107_cast_fp16 = squeeze(axes = var_107_axes_0, x = var_106_cast_fp16)[name = tensor<string, []>("op_107_cast_fp16")];
tensor<int32, [4]> q_1_begin_0 = const()[name = tensor<string, []>("q_1_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
tensor<int32, [4]> q_1_end_0 = const()[name = tensor<string, []>("q_1_end_0"), val = tensor<int32, [4]>([1, 77, 1, 512])];
tensor<bool, [4]> q_1_end_mask_0 = const()[name = tensor<string, []>("q_1_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> q_1_squeeze_mask_0 = const()[name = tensor<string, []>("q_1_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> q_1_cast_fp16 = slice_by_index(begin = q_1_begin_0, end = q_1_end_0, end_mask = q_1_end_mask_0, squeeze_mask = q_1_squeeze_mask_0, x = var_107_cast_fp16)[name = tensor<string, []>("q_1_cast_fp16")];
tensor<int32, [4]> k_1_begin_0 = const()[name = tensor<string, []>("k_1_begin_0"), val = tensor<int32, [4]>([1, 0, 0, 0])];
tensor<int32, [4]> k_1_end_0 = const()[name = tensor<string, []>("k_1_end_0"), val = tensor<int32, [4]>([2, 77, 1, 512])];
tensor<bool, [4]> k_1_end_mask_0 = const()[name = tensor<string, []>("k_1_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> k_1_squeeze_mask_0 = const()[name = tensor<string, []>("k_1_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> k_1_cast_fp16 = slice_by_index(begin = k_1_begin_0, end = k_1_end_0, end_mask = k_1_end_mask_0, squeeze_mask = k_1_squeeze_mask_0, x = var_107_cast_fp16)[name = tensor<string, []>("k_1_cast_fp16")];
tensor<int32, [4]> v_1_begin_0 = const()[name = tensor<string, []>("v_1_begin_0"), val = tensor<int32, [4]>([2, 0, 0, 0])];
tensor<int32, [4]> v_1_end_0 = const()[name = tensor<string, []>("v_1_end_0"), val = tensor<int32, [4]>([3, 77, 1, 512])];
tensor<bool, [4]> v_1_end_mask_0 = const()[name = tensor<string, []>("v_1_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> v_1_squeeze_mask_0 = const()[name = tensor<string, []>("v_1_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> v_1_cast_fp16 = slice_by_index(begin = v_1_begin_0, end = v_1_end_0, end_mask = v_1_end_mask_0, squeeze_mask = v_1_squeeze_mask_0, x = var_107_cast_fp16)[name = tensor<string, []>("v_1_cast_fp16")];
tensor<int32, [3]> var_116 = const()[name = tensor<string, []>("op_116"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_117_cast_fp16 = reshape(shape = var_116, x = q_1_cast_fp16)[name = tensor<string, []>("op_117_cast_fp16")];
tensor<int32, [3]> q_3_perm_0 = const()[name = tensor<string, []>("q_3_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_123 = const()[name = tensor<string, []>("op_123"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_124_cast_fp16 = reshape(shape = var_123, x = k_1_cast_fp16)[name = tensor<string, []>("op_124_cast_fp16")];
tensor<int32, [3]> k_3_perm_0 = const()[name = tensor<string, []>("k_3_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_130 = const()[name = tensor<string, []>("op_130"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_131_cast_fp16 = reshape(shape = var_130, x = v_1_cast_fp16)[name = tensor<string, []>("op_131_cast_fp16")];
tensor<int32, [3]> v_3_perm_0 = const()[name = tensor<string, []>("v_3_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [4]> var_135 = const()[name = tensor<string, []>("op_135"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> q_3_cast_fp16 = transpose(perm = q_3_perm_0, x = var_117_cast_fp16)[name = tensor<string, []>("transpose_83")];
tensor<fp16, [1, 8, 77, 64]> q_5_cast_fp16 = reshape(shape = var_135, x = q_3_cast_fp16)[name = tensor<string, []>("q_5_cast_fp16")];
tensor<int32, [4]> var_137 = const()[name = tensor<string, []>("op_137"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> k_3_cast_fp16 = transpose(perm = k_3_perm_0, x = var_124_cast_fp16)[name = tensor<string, []>("transpose_82")];
tensor<fp16, [1, 8, 77, 64]> k_5_cast_fp16 = reshape(shape = var_137, x = k_3_cast_fp16)[name = tensor<string, []>("k_5_cast_fp16")];
tensor<int32, [4]> var_139 = const()[name = tensor<string, []>("op_139"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> v_3_cast_fp16 = transpose(perm = v_3_perm_0, x = var_131_cast_fp16)[name = tensor<string, []>("transpose_81")];
tensor<fp16, [1, 8, 77, 64]> v_5_cast_fp16 = reshape(shape = var_139, x = v_3_cast_fp16)[name = tensor<string, []>("v_5_cast_fp16")];
tensor<fp16, []> mul_1_y_0_to_fp16 = const()[name = tensor<string, []>("mul_1_y_0_to_fp16"), val = tensor<fp16, []>(0x1p-3)];
tensor<fp16, [1, 8, 77, 64]> mul_1_cast_fp16 = mul(x = q_5_cast_fp16, y = mul_1_y_0_to_fp16)[name = tensor<string, []>("mul_1_cast_fp16")];
tensor<bool, []> matmul_0_transpose_y_0 = const()[name = tensor<string, []>("matmul_0_transpose_y_0"), val = tensor<bool, []>(true)];
tensor<bool, []> matmul_0_transpose_x_0 = const()[name = tensor<string, []>("matmul_0_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 77]> matmul_0_cast_fp16 = matmul(transpose_x = matmul_0_transpose_x_0, transpose_y = matmul_0_transpose_y_0, x = mul_1_cast_fp16, y = k_5_cast_fp16)[name = tensor<string, []>("matmul_0_cast_fp16")];
tensor<fp16, [1, 1, 77, 77]> attn_mask_7_to_fp16 = const()[name = tensor<string, []>("attn_mask_7_to_fp16"), val = tensor<fp16, [1, 1, 77, 77]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(52251072)))];
tensor<fp16, [1, 8, 77, 77]> add_0_cast_fp16 = add(x = matmul_0_cast_fp16, y = attn_mask_7_to_fp16)[name = tensor<string, []>("add_0_cast_fp16")];
tensor<int32, []> softmax_0_axis_0 = const()[name = tensor<string, []>("softmax_0_axis_0"), val = tensor<int32, []>(-1)];
tensor<fp16, [1, 8, 77, 77]> softmax_0_cast_fp16 = softmax(axis = softmax_0_axis_0, x = add_0_cast_fp16)[name = tensor<string, []>("softmax_0_cast_fp16")];
tensor<bool, []> attn_output_1_transpose_x_0 = const()[name = tensor<string, []>("attn_output_1_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<bool, []> attn_output_1_transpose_y_0 = const()[name = tensor<string, []>("attn_output_1_transpose_y_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 64]> attn_output_1_cast_fp16 = matmul(transpose_x = attn_output_1_transpose_x_0, transpose_y = attn_output_1_transpose_y_0, x = softmax_0_cast_fp16, y = v_5_cast_fp16)[name = tensor<string, []>("attn_output_1_cast_fp16")];
tensor<int32, [4]> var_142 = const()[name = tensor<string, []>("op_142"), val = tensor<int32, [4]>([2, 0, 1, 3])];
tensor<int32, [2]> var_147 = const()[name = tensor<string, []>("op_147"), val = tensor<int32, [2]>([77, 512])];
tensor<fp16, [77, 1, 8, 64]> var_143_cast_fp16 = transpose(perm = var_142, x = attn_output_1_cast_fp16)[name = tensor<string, []>("transpose_80")];
tensor<fp16, [77, 512]> attn_output_3_cast_fp16 = reshape(shape = var_147, x = var_143_cast_fp16)[name = tensor<string, []>("attn_output_3_cast_fp16")];
tensor<fp16, [512, 512]> model_transformer_resblocks_0_attn_out_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_0_attn_out_proj_weight_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(52263040)))];
tensor<fp16, [512]> model_transformer_resblocks_0_attn_out_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_0_attn_out_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(52787392)))];
tensor<fp16, [77, 512]> linear_1_cast_fp16 = linear(bias = model_transformer_resblocks_0_attn_out_proj_bias_to_fp16, weight = model_transformer_resblocks_0_attn_out_proj_weight_to_fp16, x = attn_output_3_cast_fp16)[name = tensor<string, []>("linear_1_cast_fp16")];
tensor<int32, [3]> var_151 = const()[name = tensor<string, []>("op_151"), val = tensor<int32, [3]>([77, 1, 512])];
tensor<fp16, [77, 1, 512]> attn_output_7_cast_fp16 = reshape(shape = var_151, x = linear_1_cast_fp16)[name = tensor<string, []>("attn_output_7_cast_fp16")];
tensor<int32, [3]> var_153_perm_0 = const()[name = tensor<string, []>("op_153_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1, 77, 512]> var_153_cast_fp16 = transpose(perm = var_153_perm_0, x = attn_output_7_cast_fp16)[name = tensor<string, []>("transpose_79")];
tensor<fp16, [1, 77, 512]> input_3_cast_fp16 = add(x = input_1_cast_fp16, y = var_153_cast_fp16)[name = tensor<string, []>("input_3_cast_fp16")];
tensor<int32, [1]> x_5_axes_0 = const()[name = tensor<string, []>("x_5_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_0_ln_2_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_0_ln_2_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(52788480)))];
tensor<fp16, [512]> model_transformer_resblocks_0_ln_2_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_0_ln_2_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(52789568)))];
tensor<fp16, [1, 77, 512]> x_5_cast_fp16 = layer_norm(axes = x_5_axes_0, beta = model_transformer_resblocks_0_ln_2_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_0_ln_2_weight_to_fp16, x = input_3_cast_fp16)[name = tensor<string, []>("x_5_cast_fp16")];
tensor<fp16, [2048, 512]> model_transformer_resblocks_0_mlp_c_fc_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_0_mlp_c_fc_weight_to_fp16"), val = tensor<fp16, [2048, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(52790656)))];
tensor<fp16, [2048]> model_transformer_resblocks_0_mlp_c_fc_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_0_mlp_c_fc_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(54887872)))];
tensor<fp16, [1, 77, 2048]> linear_2_cast_fp16 = linear(bias = model_transformer_resblocks_0_mlp_c_fc_bias_to_fp16, weight = model_transformer_resblocks_0_mlp_c_fc_weight_to_fp16, x = x_5_cast_fp16)[name = tensor<string, []>("linear_2_cast_fp16")];
tensor<string, []> input_9_mode_0 = const()[name = tensor<string, []>("input_9_mode_0"), val = tensor<string, []>("EXACT")];
tensor<fp16, [1, 77, 2048]> input_9_cast_fp16 = gelu(mode = input_9_mode_0, x = linear_2_cast_fp16)[name = tensor<string, []>("input_9_cast_fp16")];
tensor<fp16, [512, 2048]> model_transformer_resblocks_0_mlp_c_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_0_mlp_c_proj_weight_to_fp16"), val = tensor<fp16, [512, 2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(54892032)))];
tensor<fp16, [512]> model_transformer_resblocks_0_mlp_c_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_0_mlp_c_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(56989248)))];
tensor<fp16, [1, 77, 512]> linear_3_cast_fp16 = linear(bias = model_transformer_resblocks_0_mlp_c_proj_bias_to_fp16, weight = model_transformer_resblocks_0_mlp_c_proj_weight_to_fp16, x = input_9_cast_fp16)[name = tensor<string, []>("linear_3_cast_fp16")];
tensor<fp16, [1, 77, 512]> input_11_cast_fp16 = add(x = input_3_cast_fp16, y = linear_3_cast_fp16)[name = tensor<string, []>("input_11_cast_fp16")];
tensor<int32, [1]> x_7_axes_0 = const()[name = tensor<string, []>("x_7_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_1_ln_1_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_1_ln_1_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(56990336)))];
tensor<fp16, [512]> model_transformer_resblocks_1_ln_1_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_1_ln_1_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(56991424)))];
tensor<fp16, [1, 77, 512]> x_7_cast_fp16 = layer_norm(axes = x_7_axes_0, beta = model_transformer_resblocks_1_ln_1_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_1_ln_1_weight_to_fp16, x = input_11_cast_fp16)[name = tensor<string, []>("x_7_cast_fp16")];
tensor<int32, [3]> query_7_perm_0 = const()[name = tensor<string, []>("query_7_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1536, 512]> model_transformer_resblocks_1_attn_in_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_1_attn_in_proj_weight_to_fp16"), val = tensor<fp16, [1536, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(56992512)))];
tensor<fp16, [1536]> model_transformer_resblocks_1_attn_in_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_1_attn_in_proj_bias_to_fp16"), val = tensor<fp16, [1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(58565440)))];
tensor<fp16, [77, 1, 512]> query_7_cast_fp16 = transpose(perm = query_7_perm_0, x = x_7_cast_fp16)[name = tensor<string, []>("transpose_78")];
tensor<fp16, [77, 1, 1536]> linear_4_cast_fp16 = linear(bias = model_transformer_resblocks_1_attn_in_proj_bias_to_fp16, weight = model_transformer_resblocks_1_attn_in_proj_weight_to_fp16, x = query_7_cast_fp16)[name = tensor<string, []>("linear_4_cast_fp16")];
tensor<int32, [4]> concat_1 = const()[name = tensor<string, []>("concat_1"), val = tensor<int32, [4]>([77, 1, 3, 512])];
tensor<fp16, [77, 1, 3, 512]> var_206_cast_fp16 = reshape(shape = concat_1, x = linear_4_cast_fp16)[name = tensor<string, []>("op_206_cast_fp16")];
tensor<int32, [1]> var_207_axes_0 = const()[name = tensor<string, []>("op_207_axes_0"), val = tensor<int32, [1]>([0])];
tensor<fp16, [1, 77, 1, 3, 512]> var_207_cast_fp16 = expand_dims(axes = var_207_axes_0, x = var_206_cast_fp16)[name = tensor<string, []>("op_207_cast_fp16")];
tensor<int32, [5]> var_208_perm_0 = const()[name = tensor<string, []>("op_208_perm_0"), val = tensor<int32, [5]>([-2, 1, 2, 0, 4])];
tensor<int32, [1]> var_209_axes_0 = const()[name = tensor<string, []>("op_209_axes_0"), val = tensor<int32, [1]>([-2])];
tensor<fp16, [3, 77, 1, 1, 512]> var_208_cast_fp16 = transpose(perm = var_208_perm_0, x = var_207_cast_fp16)[name = tensor<string, []>("transpose_77")];
tensor<fp16, [3, 77, 1, 512]> var_209_cast_fp16 = squeeze(axes = var_209_axes_0, x = var_208_cast_fp16)[name = tensor<string, []>("op_209_cast_fp16")];
tensor<int32, [4]> q_7_begin_0 = const()[name = tensor<string, []>("q_7_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
tensor<int32, [4]> q_7_end_0 = const()[name = tensor<string, []>("q_7_end_0"), val = tensor<int32, [4]>([1, 77, 1, 512])];
tensor<bool, [4]> q_7_end_mask_0 = const()[name = tensor<string, []>("q_7_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> q_7_squeeze_mask_0 = const()[name = tensor<string, []>("q_7_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> q_7_cast_fp16 = slice_by_index(begin = q_7_begin_0, end = q_7_end_0, end_mask = q_7_end_mask_0, squeeze_mask = q_7_squeeze_mask_0, x = var_209_cast_fp16)[name = tensor<string, []>("q_7_cast_fp16")];
tensor<int32, [4]> k_7_begin_0 = const()[name = tensor<string, []>("k_7_begin_0"), val = tensor<int32, [4]>([1, 0, 0, 0])];
tensor<int32, [4]> k_7_end_0 = const()[name = tensor<string, []>("k_7_end_0"), val = tensor<int32, [4]>([2, 77, 1, 512])];
tensor<bool, [4]> k_7_end_mask_0 = const()[name = tensor<string, []>("k_7_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> k_7_squeeze_mask_0 = const()[name = tensor<string, []>("k_7_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> k_7_cast_fp16 = slice_by_index(begin = k_7_begin_0, end = k_7_end_0, end_mask = k_7_end_mask_0, squeeze_mask = k_7_squeeze_mask_0, x = var_209_cast_fp16)[name = tensor<string, []>("k_7_cast_fp16")];
tensor<int32, [4]> v_7_begin_0 = const()[name = tensor<string, []>("v_7_begin_0"), val = tensor<int32, [4]>([2, 0, 0, 0])];
tensor<int32, [4]> v_7_end_0 = const()[name = tensor<string, []>("v_7_end_0"), val = tensor<int32, [4]>([3, 77, 1, 512])];
tensor<bool, [4]> v_7_end_mask_0 = const()[name = tensor<string, []>("v_7_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> v_7_squeeze_mask_0 = const()[name = tensor<string, []>("v_7_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> v_7_cast_fp16 = slice_by_index(begin = v_7_begin_0, end = v_7_end_0, end_mask = v_7_end_mask_0, squeeze_mask = v_7_squeeze_mask_0, x = var_209_cast_fp16)[name = tensor<string, []>("v_7_cast_fp16")];
tensor<int32, [3]> var_218 = const()[name = tensor<string, []>("op_218"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_219_cast_fp16 = reshape(shape = var_218, x = q_7_cast_fp16)[name = tensor<string, []>("op_219_cast_fp16")];
tensor<int32, [3]> q_9_perm_0 = const()[name = tensor<string, []>("q_9_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_225 = const()[name = tensor<string, []>("op_225"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_226_cast_fp16 = reshape(shape = var_225, x = k_7_cast_fp16)[name = tensor<string, []>("op_226_cast_fp16")];
tensor<int32, [3]> k_9_perm_0 = const()[name = tensor<string, []>("k_9_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_232 = const()[name = tensor<string, []>("op_232"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_233_cast_fp16 = reshape(shape = var_232, x = v_7_cast_fp16)[name = tensor<string, []>("op_233_cast_fp16")];
tensor<int32, [3]> v_9_perm_0 = const()[name = tensor<string, []>("v_9_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [4]> var_237 = const()[name = tensor<string, []>("op_237"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> q_9_cast_fp16 = transpose(perm = q_9_perm_0, x = var_219_cast_fp16)[name = tensor<string, []>("transpose_76")];
tensor<fp16, [1, 8, 77, 64]> q_11_cast_fp16 = reshape(shape = var_237, x = q_9_cast_fp16)[name = tensor<string, []>("q_11_cast_fp16")];
tensor<int32, [4]> var_239 = const()[name = tensor<string, []>("op_239"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> k_9_cast_fp16 = transpose(perm = k_9_perm_0, x = var_226_cast_fp16)[name = tensor<string, []>("transpose_75")];
tensor<fp16, [1, 8, 77, 64]> k_11_cast_fp16 = reshape(shape = var_239, x = k_9_cast_fp16)[name = tensor<string, []>("k_11_cast_fp16")];
tensor<int32, [4]> var_241 = const()[name = tensor<string, []>("op_241"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> v_9_cast_fp16 = transpose(perm = v_9_perm_0, x = var_233_cast_fp16)[name = tensor<string, []>("transpose_74")];
tensor<fp16, [1, 8, 77, 64]> v_11_cast_fp16 = reshape(shape = var_241, x = v_9_cast_fp16)[name = tensor<string, []>("v_11_cast_fp16")];
tensor<fp16, []> mul_3_y_0_to_fp16 = const()[name = tensor<string, []>("mul_3_y_0_to_fp16"), val = tensor<fp16, []>(0x1p-3)];
tensor<fp16, [1, 8, 77, 64]> mul_3_cast_fp16 = mul(x = q_11_cast_fp16, y = mul_3_y_0_to_fp16)[name = tensor<string, []>("mul_3_cast_fp16")];
tensor<bool, []> matmul_1_transpose_y_0 = const()[name = tensor<string, []>("matmul_1_transpose_y_0"), val = tensor<bool, []>(true)];
tensor<bool, []> matmul_1_transpose_x_0 = const()[name = tensor<string, []>("matmul_1_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 77]> matmul_1_cast_fp16 = matmul(transpose_x = matmul_1_transpose_x_0, transpose_y = matmul_1_transpose_y_0, x = mul_3_cast_fp16, y = k_11_cast_fp16)[name = tensor<string, []>("matmul_1_cast_fp16")];
tensor<fp16, [1, 8, 77, 77]> add_1_cast_fp16 = add(x = matmul_1_cast_fp16, y = attn_mask_7_to_fp16)[name = tensor<string, []>("add_1_cast_fp16")];
tensor<int32, []> softmax_1_axis_0 = const()[name = tensor<string, []>("softmax_1_axis_0"), val = tensor<int32, []>(-1)];
tensor<fp16, [1, 8, 77, 77]> softmax_1_cast_fp16 = softmax(axis = softmax_1_axis_0, x = add_1_cast_fp16)[name = tensor<string, []>("softmax_1_cast_fp16")];
tensor<bool, []> attn_output_9_transpose_x_0 = const()[name = tensor<string, []>("attn_output_9_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<bool, []> attn_output_9_transpose_y_0 = const()[name = tensor<string, []>("attn_output_9_transpose_y_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 64]> attn_output_9_cast_fp16 = matmul(transpose_x = attn_output_9_transpose_x_0, transpose_y = attn_output_9_transpose_y_0, x = softmax_1_cast_fp16, y = v_11_cast_fp16)[name = tensor<string, []>("attn_output_9_cast_fp16")];
tensor<int32, [4]> var_244 = const()[name = tensor<string, []>("op_244"), val = tensor<int32, [4]>([2, 0, 1, 3])];
tensor<int32, [2]> var_249 = const()[name = tensor<string, []>("op_249"), val = tensor<int32, [2]>([77, 512])];
tensor<fp16, [77, 1, 8, 64]> var_245_cast_fp16 = transpose(perm = var_244, x = attn_output_9_cast_fp16)[name = tensor<string, []>("transpose_73")];
tensor<fp16, [77, 512]> attn_output_11_cast_fp16 = reshape(shape = var_249, x = var_245_cast_fp16)[name = tensor<string, []>("attn_output_11_cast_fp16")];
tensor<fp16, [512, 512]> model_transformer_resblocks_1_attn_out_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_1_attn_out_proj_weight_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(58568576)))];
tensor<fp16, [512]> model_transformer_resblocks_1_attn_out_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_1_attn_out_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(59092928)))];
tensor<fp16, [77, 512]> linear_5_cast_fp16 = linear(bias = model_transformer_resblocks_1_attn_out_proj_bias_to_fp16, weight = model_transformer_resblocks_1_attn_out_proj_weight_to_fp16, x = attn_output_11_cast_fp16)[name = tensor<string, []>("linear_5_cast_fp16")];
tensor<int32, [3]> var_253 = const()[name = tensor<string, []>("op_253"), val = tensor<int32, [3]>([77, 1, 512])];
tensor<fp16, [77, 1, 512]> attn_output_15_cast_fp16 = reshape(shape = var_253, x = linear_5_cast_fp16)[name = tensor<string, []>("attn_output_15_cast_fp16")];
tensor<int32, [3]> var_255_perm_0 = const()[name = tensor<string, []>("op_255_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1, 77, 512]> var_255_cast_fp16 = transpose(perm = var_255_perm_0, x = attn_output_15_cast_fp16)[name = tensor<string, []>("transpose_72")];
tensor<fp16, [1, 77, 512]> input_13_cast_fp16 = add(x = input_11_cast_fp16, y = var_255_cast_fp16)[name = tensor<string, []>("input_13_cast_fp16")];
tensor<int32, [1]> x_9_axes_0 = const()[name = tensor<string, []>("x_9_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_1_ln_2_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_1_ln_2_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(59094016)))];
tensor<fp16, [512]> model_transformer_resblocks_1_ln_2_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_1_ln_2_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(59095104)))];
tensor<fp16, [1, 77, 512]> x_9_cast_fp16 = layer_norm(axes = x_9_axes_0, beta = model_transformer_resblocks_1_ln_2_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_1_ln_2_weight_to_fp16, x = input_13_cast_fp16)[name = tensor<string, []>("x_9_cast_fp16")];
tensor<fp16, [2048, 512]> model_transformer_resblocks_1_mlp_c_fc_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_1_mlp_c_fc_weight_to_fp16"), val = tensor<fp16, [2048, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(59096192)))];
tensor<fp16, [2048]> model_transformer_resblocks_1_mlp_c_fc_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_1_mlp_c_fc_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(61193408)))];
tensor<fp16, [1, 77, 2048]> linear_6_cast_fp16 = linear(bias = model_transformer_resblocks_1_mlp_c_fc_bias_to_fp16, weight = model_transformer_resblocks_1_mlp_c_fc_weight_to_fp16, x = x_9_cast_fp16)[name = tensor<string, []>("linear_6_cast_fp16")];
tensor<string, []> input_19_mode_0 = const()[name = tensor<string, []>("input_19_mode_0"), val = tensor<string, []>("EXACT")];
tensor<fp16, [1, 77, 2048]> input_19_cast_fp16 = gelu(mode = input_19_mode_0, x = linear_6_cast_fp16)[name = tensor<string, []>("input_19_cast_fp16")];
tensor<fp16, [512, 2048]> model_transformer_resblocks_1_mlp_c_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_1_mlp_c_proj_weight_to_fp16"), val = tensor<fp16, [512, 2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(61197568)))];
tensor<fp16, [512]> model_transformer_resblocks_1_mlp_c_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_1_mlp_c_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(63294784)))];
tensor<fp16, [1, 77, 512]> linear_7_cast_fp16 = linear(bias = model_transformer_resblocks_1_mlp_c_proj_bias_to_fp16, weight = model_transformer_resblocks_1_mlp_c_proj_weight_to_fp16, x = input_19_cast_fp16)[name = tensor<string, []>("linear_7_cast_fp16")];
tensor<fp16, [1, 77, 512]> input_21_cast_fp16 = add(x = input_13_cast_fp16, y = linear_7_cast_fp16)[name = tensor<string, []>("input_21_cast_fp16")];
tensor<int32, [1]> x_11_axes_0 = const()[name = tensor<string, []>("x_11_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_2_ln_1_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_2_ln_1_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(63295872)))];
tensor<fp16, [512]> model_transformer_resblocks_2_ln_1_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_2_ln_1_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(63296960)))];
tensor<fp16, [1, 77, 512]> x_11_cast_fp16 = layer_norm(axes = x_11_axes_0, beta = model_transformer_resblocks_2_ln_1_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_2_ln_1_weight_to_fp16, x = input_21_cast_fp16)[name = tensor<string, []>("x_11_cast_fp16")];
tensor<int32, [3]> query_11_perm_0 = const()[name = tensor<string, []>("query_11_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1536, 512]> model_transformer_resblocks_2_attn_in_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_2_attn_in_proj_weight_to_fp16"), val = tensor<fp16, [1536, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(63298048)))];
tensor<fp16, [1536]> model_transformer_resblocks_2_attn_in_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_2_attn_in_proj_bias_to_fp16"), val = tensor<fp16, [1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(64870976)))];
tensor<fp16, [77, 1, 512]> query_11_cast_fp16 = transpose(perm = query_11_perm_0, x = x_11_cast_fp16)[name = tensor<string, []>("transpose_71")];
tensor<fp16, [77, 1, 1536]> linear_8_cast_fp16 = linear(bias = model_transformer_resblocks_2_attn_in_proj_bias_to_fp16, weight = model_transformer_resblocks_2_attn_in_proj_weight_to_fp16, x = query_11_cast_fp16)[name = tensor<string, []>("linear_8_cast_fp16")];
tensor<int32, [4]> concat_2 = const()[name = tensor<string, []>("concat_2"), val = tensor<int32, [4]>([77, 1, 3, 512])];
tensor<fp16, [77, 1, 3, 512]> var_308_cast_fp16 = reshape(shape = concat_2, x = linear_8_cast_fp16)[name = tensor<string, []>("op_308_cast_fp16")];
tensor<int32, [1]> var_309_axes_0 = const()[name = tensor<string, []>("op_309_axes_0"), val = tensor<int32, [1]>([0])];
tensor<fp16, [1, 77, 1, 3, 512]> var_309_cast_fp16 = expand_dims(axes = var_309_axes_0, x = var_308_cast_fp16)[name = tensor<string, []>("op_309_cast_fp16")];
tensor<int32, [5]> var_310_perm_0 = const()[name = tensor<string, []>("op_310_perm_0"), val = tensor<int32, [5]>([-2, 1, 2, 0, 4])];
tensor<int32, [1]> var_311_axes_0 = const()[name = tensor<string, []>("op_311_axes_0"), val = tensor<int32, [1]>([-2])];
tensor<fp16, [3, 77, 1, 1, 512]> var_310_cast_fp16 = transpose(perm = var_310_perm_0, x = var_309_cast_fp16)[name = tensor<string, []>("transpose_70")];
tensor<fp16, [3, 77, 1, 512]> var_311_cast_fp16 = squeeze(axes = var_311_axes_0, x = var_310_cast_fp16)[name = tensor<string, []>("op_311_cast_fp16")];
tensor<int32, [4]> q_13_begin_0 = const()[name = tensor<string, []>("q_13_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
tensor<int32, [4]> q_13_end_0 = const()[name = tensor<string, []>("q_13_end_0"), val = tensor<int32, [4]>([1, 77, 1, 512])];
tensor<bool, [4]> q_13_end_mask_0 = const()[name = tensor<string, []>("q_13_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> q_13_squeeze_mask_0 = const()[name = tensor<string, []>("q_13_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> q_13_cast_fp16 = slice_by_index(begin = q_13_begin_0, end = q_13_end_0, end_mask = q_13_end_mask_0, squeeze_mask = q_13_squeeze_mask_0, x = var_311_cast_fp16)[name = tensor<string, []>("q_13_cast_fp16")];
tensor<int32, [4]> k_13_begin_0 = const()[name = tensor<string, []>("k_13_begin_0"), val = tensor<int32, [4]>([1, 0, 0, 0])];
tensor<int32, [4]> k_13_end_0 = const()[name = tensor<string, []>("k_13_end_0"), val = tensor<int32, [4]>([2, 77, 1, 512])];
tensor<bool, [4]> k_13_end_mask_0 = const()[name = tensor<string, []>("k_13_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> k_13_squeeze_mask_0 = const()[name = tensor<string, []>("k_13_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> k_13_cast_fp16 = slice_by_index(begin = k_13_begin_0, end = k_13_end_0, end_mask = k_13_end_mask_0, squeeze_mask = k_13_squeeze_mask_0, x = var_311_cast_fp16)[name = tensor<string, []>("k_13_cast_fp16")];
tensor<int32, [4]> v_13_begin_0 = const()[name = tensor<string, []>("v_13_begin_0"), val = tensor<int32, [4]>([2, 0, 0, 0])];
tensor<int32, [4]> v_13_end_0 = const()[name = tensor<string, []>("v_13_end_0"), val = tensor<int32, [4]>([3, 77, 1, 512])];
tensor<bool, [4]> v_13_end_mask_0 = const()[name = tensor<string, []>("v_13_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> v_13_squeeze_mask_0 = const()[name = tensor<string, []>("v_13_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> v_13_cast_fp16 = slice_by_index(begin = v_13_begin_0, end = v_13_end_0, end_mask = v_13_end_mask_0, squeeze_mask = v_13_squeeze_mask_0, x = var_311_cast_fp16)[name = tensor<string, []>("v_13_cast_fp16")];
tensor<int32, [3]> var_320 = const()[name = tensor<string, []>("op_320"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_321_cast_fp16 = reshape(shape = var_320, x = q_13_cast_fp16)[name = tensor<string, []>("op_321_cast_fp16")];
tensor<int32, [3]> q_15_perm_0 = const()[name = tensor<string, []>("q_15_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_327 = const()[name = tensor<string, []>("op_327"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_328_cast_fp16 = reshape(shape = var_327, x = k_13_cast_fp16)[name = tensor<string, []>("op_328_cast_fp16")];
tensor<int32, [3]> k_15_perm_0 = const()[name = tensor<string, []>("k_15_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_334 = const()[name = tensor<string, []>("op_334"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_335_cast_fp16 = reshape(shape = var_334, x = v_13_cast_fp16)[name = tensor<string, []>("op_335_cast_fp16")];
tensor<int32, [3]> v_15_perm_0 = const()[name = tensor<string, []>("v_15_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [4]> var_339 = const()[name = tensor<string, []>("op_339"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> q_15_cast_fp16 = transpose(perm = q_15_perm_0, x = var_321_cast_fp16)[name = tensor<string, []>("transpose_69")];
tensor<fp16, [1, 8, 77, 64]> q_17_cast_fp16 = reshape(shape = var_339, x = q_15_cast_fp16)[name = tensor<string, []>("q_17_cast_fp16")];
tensor<int32, [4]> var_341 = const()[name = tensor<string, []>("op_341"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> k_15_cast_fp16 = transpose(perm = k_15_perm_0, x = var_328_cast_fp16)[name = tensor<string, []>("transpose_68")];
tensor<fp16, [1, 8, 77, 64]> k_17_cast_fp16 = reshape(shape = var_341, x = k_15_cast_fp16)[name = tensor<string, []>("k_17_cast_fp16")];
tensor<int32, [4]> var_343 = const()[name = tensor<string, []>("op_343"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> v_15_cast_fp16 = transpose(perm = v_15_perm_0, x = var_335_cast_fp16)[name = tensor<string, []>("transpose_67")];
tensor<fp16, [1, 8, 77, 64]> v_17_cast_fp16 = reshape(shape = var_343, x = v_15_cast_fp16)[name = tensor<string, []>("v_17_cast_fp16")];
tensor<fp16, []> mul_5_y_0_to_fp16 = const()[name = tensor<string, []>("mul_5_y_0_to_fp16"), val = tensor<fp16, []>(0x1p-3)];
tensor<fp16, [1, 8, 77, 64]> mul_5_cast_fp16 = mul(x = q_17_cast_fp16, y = mul_5_y_0_to_fp16)[name = tensor<string, []>("mul_5_cast_fp16")];
tensor<bool, []> matmul_2_transpose_y_0 = const()[name = tensor<string, []>("matmul_2_transpose_y_0"), val = tensor<bool, []>(true)];
tensor<bool, []> matmul_2_transpose_x_0 = const()[name = tensor<string, []>("matmul_2_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 77]> matmul_2_cast_fp16 = matmul(transpose_x = matmul_2_transpose_x_0, transpose_y = matmul_2_transpose_y_0, x = mul_5_cast_fp16, y = k_17_cast_fp16)[name = tensor<string, []>("matmul_2_cast_fp16")];
tensor<fp16, [1, 8, 77, 77]> add_2_cast_fp16 = add(x = matmul_2_cast_fp16, y = attn_mask_7_to_fp16)[name = tensor<string, []>("add_2_cast_fp16")];
tensor<int32, []> softmax_2_axis_0 = const()[name = tensor<string, []>("softmax_2_axis_0"), val = tensor<int32, []>(-1)];
tensor<fp16, [1, 8, 77, 77]> softmax_2_cast_fp16 = softmax(axis = softmax_2_axis_0, x = add_2_cast_fp16)[name = tensor<string, []>("softmax_2_cast_fp16")];
tensor<bool, []> attn_output_17_transpose_x_0 = const()[name = tensor<string, []>("attn_output_17_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<bool, []> attn_output_17_transpose_y_0 = const()[name = tensor<string, []>("attn_output_17_transpose_y_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 64]> attn_output_17_cast_fp16 = matmul(transpose_x = attn_output_17_transpose_x_0, transpose_y = attn_output_17_transpose_y_0, x = softmax_2_cast_fp16, y = v_17_cast_fp16)[name = tensor<string, []>("attn_output_17_cast_fp16")];
tensor<int32, [4]> var_346 = const()[name = tensor<string, []>("op_346"), val = tensor<int32, [4]>([2, 0, 1, 3])];
tensor<int32, [2]> var_351 = const()[name = tensor<string, []>("op_351"), val = tensor<int32, [2]>([77, 512])];
tensor<fp16, [77, 1, 8, 64]> var_347_cast_fp16 = transpose(perm = var_346, x = attn_output_17_cast_fp16)[name = tensor<string, []>("transpose_66")];
tensor<fp16, [77, 512]> attn_output_19_cast_fp16 = reshape(shape = var_351, x = var_347_cast_fp16)[name = tensor<string, []>("attn_output_19_cast_fp16")];
tensor<fp16, [512, 512]> model_transformer_resblocks_2_attn_out_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_2_attn_out_proj_weight_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(64874112)))];
tensor<fp16, [512]> model_transformer_resblocks_2_attn_out_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_2_attn_out_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(65398464)))];
tensor<fp16, [77, 512]> linear_9_cast_fp16 = linear(bias = model_transformer_resblocks_2_attn_out_proj_bias_to_fp16, weight = model_transformer_resblocks_2_attn_out_proj_weight_to_fp16, x = attn_output_19_cast_fp16)[name = tensor<string, []>("linear_9_cast_fp16")];
tensor<int32, [3]> var_355 = const()[name = tensor<string, []>("op_355"), val = tensor<int32, [3]>([77, 1, 512])];
tensor<fp16, [77, 1, 512]> attn_output_23_cast_fp16 = reshape(shape = var_355, x = linear_9_cast_fp16)[name = tensor<string, []>("attn_output_23_cast_fp16")];
tensor<int32, [3]> var_357_perm_0 = const()[name = tensor<string, []>("op_357_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1, 77, 512]> var_357_cast_fp16 = transpose(perm = var_357_perm_0, x = attn_output_23_cast_fp16)[name = tensor<string, []>("transpose_65")];
tensor<fp16, [1, 77, 512]> input_23_cast_fp16 = add(x = input_21_cast_fp16, y = var_357_cast_fp16)[name = tensor<string, []>("input_23_cast_fp16")];
tensor<int32, [1]> x_13_axes_0 = const()[name = tensor<string, []>("x_13_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_2_ln_2_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_2_ln_2_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(65399552)))];
tensor<fp16, [512]> model_transformer_resblocks_2_ln_2_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_2_ln_2_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(65400640)))];
tensor<fp16, [1, 77, 512]> x_13_cast_fp16 = layer_norm(axes = x_13_axes_0, beta = model_transformer_resblocks_2_ln_2_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_2_ln_2_weight_to_fp16, x = input_23_cast_fp16)[name = tensor<string, []>("x_13_cast_fp16")];
tensor<fp16, [2048, 512]> model_transformer_resblocks_2_mlp_c_fc_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_2_mlp_c_fc_weight_to_fp16"), val = tensor<fp16, [2048, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(65401728)))];
tensor<fp16, [2048]> model_transformer_resblocks_2_mlp_c_fc_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_2_mlp_c_fc_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(67498944)))];
tensor<fp16, [1, 77, 2048]> linear_10_cast_fp16 = linear(bias = model_transformer_resblocks_2_mlp_c_fc_bias_to_fp16, weight = model_transformer_resblocks_2_mlp_c_fc_weight_to_fp16, x = x_13_cast_fp16)[name = tensor<string, []>("linear_10_cast_fp16")];
tensor<string, []> input_29_mode_0 = const()[name = tensor<string, []>("input_29_mode_0"), val = tensor<string, []>("EXACT")];
tensor<fp16, [1, 77, 2048]> input_29_cast_fp16 = gelu(mode = input_29_mode_0, x = linear_10_cast_fp16)[name = tensor<string, []>("input_29_cast_fp16")];
tensor<fp16, [512, 2048]> model_transformer_resblocks_2_mlp_c_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_2_mlp_c_proj_weight_to_fp16"), val = tensor<fp16, [512, 2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(67503104)))];
tensor<fp16, [512]> model_transformer_resblocks_2_mlp_c_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_2_mlp_c_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(69600320)))];
tensor<fp16, [1, 77, 512]> linear_11_cast_fp16 = linear(bias = model_transformer_resblocks_2_mlp_c_proj_bias_to_fp16, weight = model_transformer_resblocks_2_mlp_c_proj_weight_to_fp16, x = input_29_cast_fp16)[name = tensor<string, []>("linear_11_cast_fp16")];
tensor<fp16, [1, 77, 512]> input_31_cast_fp16 = add(x = input_23_cast_fp16, y = linear_11_cast_fp16)[name = tensor<string, []>("input_31_cast_fp16")];
tensor<int32, [1]> x_15_axes_0 = const()[name = tensor<string, []>("x_15_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_3_ln_1_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_3_ln_1_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(69601408)))];
tensor<fp16, [512]> model_transformer_resblocks_3_ln_1_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_3_ln_1_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(69602496)))];
tensor<fp16, [1, 77, 512]> x_15_cast_fp16 = layer_norm(axes = x_15_axes_0, beta = model_transformer_resblocks_3_ln_1_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_3_ln_1_weight_to_fp16, x = input_31_cast_fp16)[name = tensor<string, []>("x_15_cast_fp16")];
tensor<int32, [3]> query_15_perm_0 = const()[name = tensor<string, []>("query_15_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1536, 512]> model_transformer_resblocks_3_attn_in_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_3_attn_in_proj_weight_to_fp16"), val = tensor<fp16, [1536, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(69603584)))];
tensor<fp16, [1536]> model_transformer_resblocks_3_attn_in_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_3_attn_in_proj_bias_to_fp16"), val = tensor<fp16, [1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(71176512)))];
tensor<fp16, [77, 1, 512]> query_15_cast_fp16 = transpose(perm = query_15_perm_0, x = x_15_cast_fp16)[name = tensor<string, []>("transpose_64")];
tensor<fp16, [77, 1, 1536]> linear_12_cast_fp16 = linear(bias = model_transformer_resblocks_3_attn_in_proj_bias_to_fp16, weight = model_transformer_resblocks_3_attn_in_proj_weight_to_fp16, x = query_15_cast_fp16)[name = tensor<string, []>("linear_12_cast_fp16")];
tensor<int32, [4]> concat_3 = const()[name = tensor<string, []>("concat_3"), val = tensor<int32, [4]>([77, 1, 3, 512])];
tensor<fp16, [77, 1, 3, 512]> var_410_cast_fp16 = reshape(shape = concat_3, x = linear_12_cast_fp16)[name = tensor<string, []>("op_410_cast_fp16")];
tensor<int32, [1]> var_411_axes_0 = const()[name = tensor<string, []>("op_411_axes_0"), val = tensor<int32, [1]>([0])];
tensor<fp16, [1, 77, 1, 3, 512]> var_411_cast_fp16 = expand_dims(axes = var_411_axes_0, x = var_410_cast_fp16)[name = tensor<string, []>("op_411_cast_fp16")];
tensor<int32, [5]> var_412_perm_0 = const()[name = tensor<string, []>("op_412_perm_0"), val = tensor<int32, [5]>([-2, 1, 2, 0, 4])];
tensor<int32, [1]> var_413_axes_0 = const()[name = tensor<string, []>("op_413_axes_0"), val = tensor<int32, [1]>([-2])];
tensor<fp16, [3, 77, 1, 1, 512]> var_412_cast_fp16 = transpose(perm = var_412_perm_0, x = var_411_cast_fp16)[name = tensor<string, []>("transpose_63")];
tensor<fp16, [3, 77, 1, 512]> var_413_cast_fp16 = squeeze(axes = var_413_axes_0, x = var_412_cast_fp16)[name = tensor<string, []>("op_413_cast_fp16")];
tensor<int32, [4]> q_19_begin_0 = const()[name = tensor<string, []>("q_19_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
tensor<int32, [4]> q_19_end_0 = const()[name = tensor<string, []>("q_19_end_0"), val = tensor<int32, [4]>([1, 77, 1, 512])];
tensor<bool, [4]> q_19_end_mask_0 = const()[name = tensor<string, []>("q_19_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> q_19_squeeze_mask_0 = const()[name = tensor<string, []>("q_19_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> q_19_cast_fp16 = slice_by_index(begin = q_19_begin_0, end = q_19_end_0, end_mask = q_19_end_mask_0, squeeze_mask = q_19_squeeze_mask_0, x = var_413_cast_fp16)[name = tensor<string, []>("q_19_cast_fp16")];
tensor<int32, [4]> k_19_begin_0 = const()[name = tensor<string, []>("k_19_begin_0"), val = tensor<int32, [4]>([1, 0, 0, 0])];
tensor<int32, [4]> k_19_end_0 = const()[name = tensor<string, []>("k_19_end_0"), val = tensor<int32, [4]>([2, 77, 1, 512])];
tensor<bool, [4]> k_19_end_mask_0 = const()[name = tensor<string, []>("k_19_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> k_19_squeeze_mask_0 = const()[name = tensor<string, []>("k_19_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> k_19_cast_fp16 = slice_by_index(begin = k_19_begin_0, end = k_19_end_0, end_mask = k_19_end_mask_0, squeeze_mask = k_19_squeeze_mask_0, x = var_413_cast_fp16)[name = tensor<string, []>("k_19_cast_fp16")];
tensor<int32, [4]> v_19_begin_0 = const()[name = tensor<string, []>("v_19_begin_0"), val = tensor<int32, [4]>([2, 0, 0, 0])];
tensor<int32, [4]> v_19_end_0 = const()[name = tensor<string, []>("v_19_end_0"), val = tensor<int32, [4]>([3, 77, 1, 512])];
tensor<bool, [4]> v_19_end_mask_0 = const()[name = tensor<string, []>("v_19_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> v_19_squeeze_mask_0 = const()[name = tensor<string, []>("v_19_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> v_19_cast_fp16 = slice_by_index(begin = v_19_begin_0, end = v_19_end_0, end_mask = v_19_end_mask_0, squeeze_mask = v_19_squeeze_mask_0, x = var_413_cast_fp16)[name = tensor<string, []>("v_19_cast_fp16")];
tensor<int32, [3]> var_422 = const()[name = tensor<string, []>("op_422"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_423_cast_fp16 = reshape(shape = var_422, x = q_19_cast_fp16)[name = tensor<string, []>("op_423_cast_fp16")];
tensor<int32, [3]> q_21_perm_0 = const()[name = tensor<string, []>("q_21_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_429 = const()[name = tensor<string, []>("op_429"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_430_cast_fp16 = reshape(shape = var_429, x = k_19_cast_fp16)[name = tensor<string, []>("op_430_cast_fp16")];
tensor<int32, [3]> k_21_perm_0 = const()[name = tensor<string, []>("k_21_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_436 = const()[name = tensor<string, []>("op_436"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_437_cast_fp16 = reshape(shape = var_436, x = v_19_cast_fp16)[name = tensor<string, []>("op_437_cast_fp16")];
tensor<int32, [3]> v_21_perm_0 = const()[name = tensor<string, []>("v_21_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [4]> var_441 = const()[name = tensor<string, []>("op_441"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> q_21_cast_fp16 = transpose(perm = q_21_perm_0, x = var_423_cast_fp16)[name = tensor<string, []>("transpose_62")];
tensor<fp16, [1, 8, 77, 64]> q_23_cast_fp16 = reshape(shape = var_441, x = q_21_cast_fp16)[name = tensor<string, []>("q_23_cast_fp16")];
tensor<int32, [4]> var_443 = const()[name = tensor<string, []>("op_443"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> k_21_cast_fp16 = transpose(perm = k_21_perm_0, x = var_430_cast_fp16)[name = tensor<string, []>("transpose_61")];
tensor<fp16, [1, 8, 77, 64]> k_23_cast_fp16 = reshape(shape = var_443, x = k_21_cast_fp16)[name = tensor<string, []>("k_23_cast_fp16")];
tensor<int32, [4]> var_445 = const()[name = tensor<string, []>("op_445"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> v_21_cast_fp16 = transpose(perm = v_21_perm_0, x = var_437_cast_fp16)[name = tensor<string, []>("transpose_60")];
tensor<fp16, [1, 8, 77, 64]> v_23_cast_fp16 = reshape(shape = var_445, x = v_21_cast_fp16)[name = tensor<string, []>("v_23_cast_fp16")];
tensor<fp16, []> mul_7_y_0_to_fp16 = const()[name = tensor<string, []>("mul_7_y_0_to_fp16"), val = tensor<fp16, []>(0x1p-3)];
tensor<fp16, [1, 8, 77, 64]> mul_7_cast_fp16 = mul(x = q_23_cast_fp16, y = mul_7_y_0_to_fp16)[name = tensor<string, []>("mul_7_cast_fp16")];
tensor<bool, []> matmul_3_transpose_y_0 = const()[name = tensor<string, []>("matmul_3_transpose_y_0"), val = tensor<bool, []>(true)];
tensor<bool, []> matmul_3_transpose_x_0 = const()[name = tensor<string, []>("matmul_3_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 77]> matmul_3_cast_fp16 = matmul(transpose_x = matmul_3_transpose_x_0, transpose_y = matmul_3_transpose_y_0, x = mul_7_cast_fp16, y = k_23_cast_fp16)[name = tensor<string, []>("matmul_3_cast_fp16")];
tensor<fp16, [1, 8, 77, 77]> add_3_cast_fp16 = add(x = matmul_3_cast_fp16, y = attn_mask_7_to_fp16)[name = tensor<string, []>("add_3_cast_fp16")];
tensor<int32, []> softmax_3_axis_0 = const()[name = tensor<string, []>("softmax_3_axis_0"), val = tensor<int32, []>(-1)];
tensor<fp16, [1, 8, 77, 77]> softmax_3_cast_fp16 = softmax(axis = softmax_3_axis_0, x = add_3_cast_fp16)[name = tensor<string, []>("softmax_3_cast_fp16")];
tensor<bool, []> attn_output_25_transpose_x_0 = const()[name = tensor<string, []>("attn_output_25_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<bool, []> attn_output_25_transpose_y_0 = const()[name = tensor<string, []>("attn_output_25_transpose_y_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 64]> attn_output_25_cast_fp16 = matmul(transpose_x = attn_output_25_transpose_x_0, transpose_y = attn_output_25_transpose_y_0, x = softmax_3_cast_fp16, y = v_23_cast_fp16)[name = tensor<string, []>("attn_output_25_cast_fp16")];
tensor<int32, [4]> var_448 = const()[name = tensor<string, []>("op_448"), val = tensor<int32, [4]>([2, 0, 1, 3])];
tensor<int32, [2]> var_453 = const()[name = tensor<string, []>("op_453"), val = tensor<int32, [2]>([77, 512])];
tensor<fp16, [77, 1, 8, 64]> var_449_cast_fp16 = transpose(perm = var_448, x = attn_output_25_cast_fp16)[name = tensor<string, []>("transpose_59")];
tensor<fp16, [77, 512]> attn_output_27_cast_fp16 = reshape(shape = var_453, x = var_449_cast_fp16)[name = tensor<string, []>("attn_output_27_cast_fp16")];
tensor<fp16, [512, 512]> model_transformer_resblocks_3_attn_out_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_3_attn_out_proj_weight_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(71179648)))];
tensor<fp16, [512]> model_transformer_resblocks_3_attn_out_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_3_attn_out_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(71704000)))];
tensor<fp16, [77, 512]> linear_13_cast_fp16 = linear(bias = model_transformer_resblocks_3_attn_out_proj_bias_to_fp16, weight = model_transformer_resblocks_3_attn_out_proj_weight_to_fp16, x = attn_output_27_cast_fp16)[name = tensor<string, []>("linear_13_cast_fp16")];
tensor<int32, [3]> var_457 = const()[name = tensor<string, []>("op_457"), val = tensor<int32, [3]>([77, 1, 512])];
tensor<fp16, [77, 1, 512]> attn_output_31_cast_fp16 = reshape(shape = var_457, x = linear_13_cast_fp16)[name = tensor<string, []>("attn_output_31_cast_fp16")];
tensor<int32, [3]> var_459_perm_0 = const()[name = tensor<string, []>("op_459_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1, 77, 512]> var_459_cast_fp16 = transpose(perm = var_459_perm_0, x = attn_output_31_cast_fp16)[name = tensor<string, []>("transpose_58")];
tensor<fp16, [1, 77, 512]> input_33_cast_fp16 = add(x = input_31_cast_fp16, y = var_459_cast_fp16)[name = tensor<string, []>("input_33_cast_fp16")];
tensor<int32, [1]> x_17_axes_0 = const()[name = tensor<string, []>("x_17_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_3_ln_2_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_3_ln_2_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(71705088)))];
tensor<fp16, [512]> model_transformer_resblocks_3_ln_2_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_3_ln_2_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(71706176)))];
tensor<fp16, [1, 77, 512]> x_17_cast_fp16 = layer_norm(axes = x_17_axes_0, beta = model_transformer_resblocks_3_ln_2_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_3_ln_2_weight_to_fp16, x = input_33_cast_fp16)[name = tensor<string, []>("x_17_cast_fp16")];
tensor<fp16, [2048, 512]> model_transformer_resblocks_3_mlp_c_fc_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_3_mlp_c_fc_weight_to_fp16"), val = tensor<fp16, [2048, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(71707264)))];
tensor<fp16, [2048]> model_transformer_resblocks_3_mlp_c_fc_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_3_mlp_c_fc_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(73804480)))];
tensor<fp16, [1, 77, 2048]> linear_14_cast_fp16 = linear(bias = model_transformer_resblocks_3_mlp_c_fc_bias_to_fp16, weight = model_transformer_resblocks_3_mlp_c_fc_weight_to_fp16, x = x_17_cast_fp16)[name = tensor<string, []>("linear_14_cast_fp16")];
tensor<string, []> input_39_mode_0 = const()[name = tensor<string, []>("input_39_mode_0"), val = tensor<string, []>("EXACT")];
tensor<fp16, [1, 77, 2048]> input_39_cast_fp16 = gelu(mode = input_39_mode_0, x = linear_14_cast_fp16)[name = tensor<string, []>("input_39_cast_fp16")];
tensor<fp16, [512, 2048]> model_transformer_resblocks_3_mlp_c_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_3_mlp_c_proj_weight_to_fp16"), val = tensor<fp16, [512, 2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(73808640)))];
tensor<fp16, [512]> model_transformer_resblocks_3_mlp_c_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_3_mlp_c_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(75905856)))];
tensor<fp16, [1, 77, 512]> linear_15_cast_fp16 = linear(bias = model_transformer_resblocks_3_mlp_c_proj_bias_to_fp16, weight = model_transformer_resblocks_3_mlp_c_proj_weight_to_fp16, x = input_39_cast_fp16)[name = tensor<string, []>("linear_15_cast_fp16")];
tensor<fp16, [1, 77, 512]> input_41_cast_fp16 = add(x = input_33_cast_fp16, y = linear_15_cast_fp16)[name = tensor<string, []>("input_41_cast_fp16")];
tensor<int32, [1]> x_19_axes_0 = const()[name = tensor<string, []>("x_19_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_4_ln_1_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_4_ln_1_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(75906944)))];
tensor<fp16, [512]> model_transformer_resblocks_4_ln_1_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_4_ln_1_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(75908032)))];
tensor<fp16, [1, 77, 512]> x_19_cast_fp16 = layer_norm(axes = x_19_axes_0, beta = model_transformer_resblocks_4_ln_1_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_4_ln_1_weight_to_fp16, x = input_41_cast_fp16)[name = tensor<string, []>("x_19_cast_fp16")];
tensor<int32, [3]> query_19_perm_0 = const()[name = tensor<string, []>("query_19_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1536, 512]> model_transformer_resblocks_4_attn_in_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_4_attn_in_proj_weight_to_fp16"), val = tensor<fp16, [1536, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(75909120)))];
tensor<fp16, [1536]> model_transformer_resblocks_4_attn_in_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_4_attn_in_proj_bias_to_fp16"), val = tensor<fp16, [1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(77482048)))];
tensor<fp16, [77, 1, 512]> query_19_cast_fp16 = transpose(perm = query_19_perm_0, x = x_19_cast_fp16)[name = tensor<string, []>("transpose_57")];
tensor<fp16, [77, 1, 1536]> linear_16_cast_fp16 = linear(bias = model_transformer_resblocks_4_attn_in_proj_bias_to_fp16, weight = model_transformer_resblocks_4_attn_in_proj_weight_to_fp16, x = query_19_cast_fp16)[name = tensor<string, []>("linear_16_cast_fp16")];
tensor<int32, [4]> concat_4 = const()[name = tensor<string, []>("concat_4"), val = tensor<int32, [4]>([77, 1, 3, 512])];
tensor<fp16, [77, 1, 3, 512]> var_512_cast_fp16 = reshape(shape = concat_4, x = linear_16_cast_fp16)[name = tensor<string, []>("op_512_cast_fp16")];
tensor<int32, [1]> var_513_axes_0 = const()[name = tensor<string, []>("op_513_axes_0"), val = tensor<int32, [1]>([0])];
tensor<fp16, [1, 77, 1, 3, 512]> var_513_cast_fp16 = expand_dims(axes = var_513_axes_0, x = var_512_cast_fp16)[name = tensor<string, []>("op_513_cast_fp16")];
tensor<int32, [5]> var_514_perm_0 = const()[name = tensor<string, []>("op_514_perm_0"), val = tensor<int32, [5]>([-2, 1, 2, 0, 4])];
tensor<int32, [1]> var_515_axes_0 = const()[name = tensor<string, []>("op_515_axes_0"), val = tensor<int32, [1]>([-2])];
tensor<fp16, [3, 77, 1, 1, 512]> var_514_cast_fp16 = transpose(perm = var_514_perm_0, x = var_513_cast_fp16)[name = tensor<string, []>("transpose_56")];
tensor<fp16, [3, 77, 1, 512]> var_515_cast_fp16 = squeeze(axes = var_515_axes_0, x = var_514_cast_fp16)[name = tensor<string, []>("op_515_cast_fp16")];
tensor<int32, [4]> q_25_begin_0 = const()[name = tensor<string, []>("q_25_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
tensor<int32, [4]> q_25_end_0 = const()[name = tensor<string, []>("q_25_end_0"), val = tensor<int32, [4]>([1, 77, 1, 512])];
tensor<bool, [4]> q_25_end_mask_0 = const()[name = tensor<string, []>("q_25_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> q_25_squeeze_mask_0 = const()[name = tensor<string, []>("q_25_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> q_25_cast_fp16 = slice_by_index(begin = q_25_begin_0, end = q_25_end_0, end_mask = q_25_end_mask_0, squeeze_mask = q_25_squeeze_mask_0, x = var_515_cast_fp16)[name = tensor<string, []>("q_25_cast_fp16")];
tensor<int32, [4]> k_25_begin_0 = const()[name = tensor<string, []>("k_25_begin_0"), val = tensor<int32, [4]>([1, 0, 0, 0])];
tensor<int32, [4]> k_25_end_0 = const()[name = tensor<string, []>("k_25_end_0"), val = tensor<int32, [4]>([2, 77, 1, 512])];
tensor<bool, [4]> k_25_end_mask_0 = const()[name = tensor<string, []>("k_25_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> k_25_squeeze_mask_0 = const()[name = tensor<string, []>("k_25_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> k_25_cast_fp16 = slice_by_index(begin = k_25_begin_0, end = k_25_end_0, end_mask = k_25_end_mask_0, squeeze_mask = k_25_squeeze_mask_0, x = var_515_cast_fp16)[name = tensor<string, []>("k_25_cast_fp16")];
tensor<int32, [4]> v_25_begin_0 = const()[name = tensor<string, []>("v_25_begin_0"), val = tensor<int32, [4]>([2, 0, 0, 0])];
tensor<int32, [4]> v_25_end_0 = const()[name = tensor<string, []>("v_25_end_0"), val = tensor<int32, [4]>([3, 77, 1, 512])];
tensor<bool, [4]> v_25_end_mask_0 = const()[name = tensor<string, []>("v_25_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> v_25_squeeze_mask_0 = const()[name = tensor<string, []>("v_25_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> v_25_cast_fp16 = slice_by_index(begin = v_25_begin_0, end = v_25_end_0, end_mask = v_25_end_mask_0, squeeze_mask = v_25_squeeze_mask_0, x = var_515_cast_fp16)[name = tensor<string, []>("v_25_cast_fp16")];
tensor<int32, [3]> var_524 = const()[name = tensor<string, []>("op_524"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_525_cast_fp16 = reshape(shape = var_524, x = q_25_cast_fp16)[name = tensor<string, []>("op_525_cast_fp16")];
tensor<int32, [3]> q_27_perm_0 = const()[name = tensor<string, []>("q_27_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_531 = const()[name = tensor<string, []>("op_531"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_532_cast_fp16 = reshape(shape = var_531, x = k_25_cast_fp16)[name = tensor<string, []>("op_532_cast_fp16")];
tensor<int32, [3]> k_27_perm_0 = const()[name = tensor<string, []>("k_27_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_538 = const()[name = tensor<string, []>("op_538"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_539_cast_fp16 = reshape(shape = var_538, x = v_25_cast_fp16)[name = tensor<string, []>("op_539_cast_fp16")];
tensor<int32, [3]> v_27_perm_0 = const()[name = tensor<string, []>("v_27_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [4]> var_543 = const()[name = tensor<string, []>("op_543"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> q_27_cast_fp16 = transpose(perm = q_27_perm_0, x = var_525_cast_fp16)[name = tensor<string, []>("transpose_55")];
tensor<fp16, [1, 8, 77, 64]> q_29_cast_fp16 = reshape(shape = var_543, x = q_27_cast_fp16)[name = tensor<string, []>("q_29_cast_fp16")];
tensor<int32, [4]> var_545 = const()[name = tensor<string, []>("op_545"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> k_27_cast_fp16 = transpose(perm = k_27_perm_0, x = var_532_cast_fp16)[name = tensor<string, []>("transpose_54")];
tensor<fp16, [1, 8, 77, 64]> k_29_cast_fp16 = reshape(shape = var_545, x = k_27_cast_fp16)[name = tensor<string, []>("k_29_cast_fp16")];
tensor<int32, [4]> var_547 = const()[name = tensor<string, []>("op_547"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> v_27_cast_fp16 = transpose(perm = v_27_perm_0, x = var_539_cast_fp16)[name = tensor<string, []>("transpose_53")];
tensor<fp16, [1, 8, 77, 64]> v_29_cast_fp16 = reshape(shape = var_547, x = v_27_cast_fp16)[name = tensor<string, []>("v_29_cast_fp16")];
tensor<fp16, []> mul_9_y_0_to_fp16 = const()[name = tensor<string, []>("mul_9_y_0_to_fp16"), val = tensor<fp16, []>(0x1p-3)];
tensor<fp16, [1, 8, 77, 64]> mul_9_cast_fp16 = mul(x = q_29_cast_fp16, y = mul_9_y_0_to_fp16)[name = tensor<string, []>("mul_9_cast_fp16")];
tensor<bool, []> matmul_4_transpose_y_0 = const()[name = tensor<string, []>("matmul_4_transpose_y_0"), val = tensor<bool, []>(true)];
tensor<bool, []> matmul_4_transpose_x_0 = const()[name = tensor<string, []>("matmul_4_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 77]> matmul_4_cast_fp16 = matmul(transpose_x = matmul_4_transpose_x_0, transpose_y = matmul_4_transpose_y_0, x = mul_9_cast_fp16, y = k_29_cast_fp16)[name = tensor<string, []>("matmul_4_cast_fp16")];
tensor<fp16, [1, 8, 77, 77]> add_4_cast_fp16 = add(x = matmul_4_cast_fp16, y = attn_mask_7_to_fp16)[name = tensor<string, []>("add_4_cast_fp16")];
tensor<int32, []> softmax_4_axis_0 = const()[name = tensor<string, []>("softmax_4_axis_0"), val = tensor<int32, []>(-1)];
tensor<fp16, [1, 8, 77, 77]> softmax_4_cast_fp16 = softmax(axis = softmax_4_axis_0, x = add_4_cast_fp16)[name = tensor<string, []>("softmax_4_cast_fp16")];
tensor<bool, []> attn_output_33_transpose_x_0 = const()[name = tensor<string, []>("attn_output_33_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<bool, []> attn_output_33_transpose_y_0 = const()[name = tensor<string, []>("attn_output_33_transpose_y_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 64]> attn_output_33_cast_fp16 = matmul(transpose_x = attn_output_33_transpose_x_0, transpose_y = attn_output_33_transpose_y_0, x = softmax_4_cast_fp16, y = v_29_cast_fp16)[name = tensor<string, []>("attn_output_33_cast_fp16")];
tensor<int32, [4]> var_550 = const()[name = tensor<string, []>("op_550"), val = tensor<int32, [4]>([2, 0, 1, 3])];
tensor<int32, [2]> var_555 = const()[name = tensor<string, []>("op_555"), val = tensor<int32, [2]>([77, 512])];
tensor<fp16, [77, 1, 8, 64]> var_551_cast_fp16 = transpose(perm = var_550, x = attn_output_33_cast_fp16)[name = tensor<string, []>("transpose_52")];
tensor<fp16, [77, 512]> attn_output_35_cast_fp16 = reshape(shape = var_555, x = var_551_cast_fp16)[name = tensor<string, []>("attn_output_35_cast_fp16")];
tensor<fp16, [512, 512]> model_transformer_resblocks_4_attn_out_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_4_attn_out_proj_weight_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(77485184)))];
tensor<fp16, [512]> model_transformer_resblocks_4_attn_out_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_4_attn_out_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(78009536)))];
tensor<fp16, [77, 512]> linear_17_cast_fp16 = linear(bias = model_transformer_resblocks_4_attn_out_proj_bias_to_fp16, weight = model_transformer_resblocks_4_attn_out_proj_weight_to_fp16, x = attn_output_35_cast_fp16)[name = tensor<string, []>("linear_17_cast_fp16")];
tensor<int32, [3]> var_559 = const()[name = tensor<string, []>("op_559"), val = tensor<int32, [3]>([77, 1, 512])];
tensor<fp16, [77, 1, 512]> attn_output_39_cast_fp16 = reshape(shape = var_559, x = linear_17_cast_fp16)[name = tensor<string, []>("attn_output_39_cast_fp16")];
tensor<int32, [3]> var_561_perm_0 = const()[name = tensor<string, []>("op_561_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1, 77, 512]> var_561_cast_fp16 = transpose(perm = var_561_perm_0, x = attn_output_39_cast_fp16)[name = tensor<string, []>("transpose_51")];
tensor<fp16, [1, 77, 512]> input_43_cast_fp16 = add(x = input_41_cast_fp16, y = var_561_cast_fp16)[name = tensor<string, []>("input_43_cast_fp16")];
tensor<int32, [1]> x_21_axes_0 = const()[name = tensor<string, []>("x_21_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_4_ln_2_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_4_ln_2_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(78010624)))];
tensor<fp16, [512]> model_transformer_resblocks_4_ln_2_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_4_ln_2_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(78011712)))];
tensor<fp16, [1, 77, 512]> x_21_cast_fp16 = layer_norm(axes = x_21_axes_0, beta = model_transformer_resblocks_4_ln_2_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_4_ln_2_weight_to_fp16, x = input_43_cast_fp16)[name = tensor<string, []>("x_21_cast_fp16")];
tensor<fp16, [2048, 512]> model_transformer_resblocks_4_mlp_c_fc_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_4_mlp_c_fc_weight_to_fp16"), val = tensor<fp16, [2048, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(78012800)))];
tensor<fp16, [2048]> model_transformer_resblocks_4_mlp_c_fc_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_4_mlp_c_fc_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(80110016)))];
tensor<fp16, [1, 77, 2048]> linear_18_cast_fp16 = linear(bias = model_transformer_resblocks_4_mlp_c_fc_bias_to_fp16, weight = model_transformer_resblocks_4_mlp_c_fc_weight_to_fp16, x = x_21_cast_fp16)[name = tensor<string, []>("linear_18_cast_fp16")];
tensor<string, []> input_49_mode_0 = const()[name = tensor<string, []>("input_49_mode_0"), val = tensor<string, []>("EXACT")];
tensor<fp16, [1, 77, 2048]> input_49_cast_fp16 = gelu(mode = input_49_mode_0, x = linear_18_cast_fp16)[name = tensor<string, []>("input_49_cast_fp16")];
tensor<fp16, [512, 2048]> model_transformer_resblocks_4_mlp_c_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_4_mlp_c_proj_weight_to_fp16"), val = tensor<fp16, [512, 2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(80114176)))];
tensor<fp16, [512]> model_transformer_resblocks_4_mlp_c_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_4_mlp_c_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(82211392)))];
tensor<fp16, [1, 77, 512]> linear_19_cast_fp16 = linear(bias = model_transformer_resblocks_4_mlp_c_proj_bias_to_fp16, weight = model_transformer_resblocks_4_mlp_c_proj_weight_to_fp16, x = input_49_cast_fp16)[name = tensor<string, []>("linear_19_cast_fp16")];
tensor<fp16, [1, 77, 512]> input_51_cast_fp16 = add(x = input_43_cast_fp16, y = linear_19_cast_fp16)[name = tensor<string, []>("input_51_cast_fp16")];
tensor<int32, [1]> x_23_axes_0 = const()[name = tensor<string, []>("x_23_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_5_ln_1_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_5_ln_1_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(82212480)))];
tensor<fp16, [512]> model_transformer_resblocks_5_ln_1_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_5_ln_1_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(82213568)))];
tensor<fp16, [1, 77, 512]> x_23_cast_fp16 = layer_norm(axes = x_23_axes_0, beta = model_transformer_resblocks_5_ln_1_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_5_ln_1_weight_to_fp16, x = input_51_cast_fp16)[name = tensor<string, []>("x_23_cast_fp16")];
tensor<int32, [3]> query_23_perm_0 = const()[name = tensor<string, []>("query_23_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1536, 512]> model_transformer_resblocks_5_attn_in_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_5_attn_in_proj_weight_to_fp16"), val = tensor<fp16, [1536, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(82214656)))];
tensor<fp16, [1536]> model_transformer_resblocks_5_attn_in_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_5_attn_in_proj_bias_to_fp16"), val = tensor<fp16, [1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(83787584)))];
tensor<fp16, [77, 1, 512]> query_23_cast_fp16 = transpose(perm = query_23_perm_0, x = x_23_cast_fp16)[name = tensor<string, []>("transpose_50")];
tensor<fp16, [77, 1, 1536]> linear_20_cast_fp16 = linear(bias = model_transformer_resblocks_5_attn_in_proj_bias_to_fp16, weight = model_transformer_resblocks_5_attn_in_proj_weight_to_fp16, x = query_23_cast_fp16)[name = tensor<string, []>("linear_20_cast_fp16")];
tensor<int32, [4]> concat_5 = const()[name = tensor<string, []>("concat_5"), val = tensor<int32, [4]>([77, 1, 3, 512])];
tensor<fp16, [77, 1, 3, 512]> var_614_cast_fp16 = reshape(shape = concat_5, x = linear_20_cast_fp16)[name = tensor<string, []>("op_614_cast_fp16")];
tensor<int32, [1]> var_615_axes_0 = const()[name = tensor<string, []>("op_615_axes_0"), val = tensor<int32, [1]>([0])];
tensor<fp16, [1, 77, 1, 3, 512]> var_615_cast_fp16 = expand_dims(axes = var_615_axes_0, x = var_614_cast_fp16)[name = tensor<string, []>("op_615_cast_fp16")];
tensor<int32, [5]> var_616_perm_0 = const()[name = tensor<string, []>("op_616_perm_0"), val = tensor<int32, [5]>([-2, 1, 2, 0, 4])];
tensor<int32, [1]> var_617_axes_0 = const()[name = tensor<string, []>("op_617_axes_0"), val = tensor<int32, [1]>([-2])];
tensor<fp16, [3, 77, 1, 1, 512]> var_616_cast_fp16 = transpose(perm = var_616_perm_0, x = var_615_cast_fp16)[name = tensor<string, []>("transpose_49")];
tensor<fp16, [3, 77, 1, 512]> var_617_cast_fp16 = squeeze(axes = var_617_axes_0, x = var_616_cast_fp16)[name = tensor<string, []>("op_617_cast_fp16")];
tensor<int32, [4]> q_31_begin_0 = const()[name = tensor<string, []>("q_31_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
tensor<int32, [4]> q_31_end_0 = const()[name = tensor<string, []>("q_31_end_0"), val = tensor<int32, [4]>([1, 77, 1, 512])];
tensor<bool, [4]> q_31_end_mask_0 = const()[name = tensor<string, []>("q_31_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> q_31_squeeze_mask_0 = const()[name = tensor<string, []>("q_31_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> q_31_cast_fp16 = slice_by_index(begin = q_31_begin_0, end = q_31_end_0, end_mask = q_31_end_mask_0, squeeze_mask = q_31_squeeze_mask_0, x = var_617_cast_fp16)[name = tensor<string, []>("q_31_cast_fp16")];
tensor<int32, [4]> k_31_begin_0 = const()[name = tensor<string, []>("k_31_begin_0"), val = tensor<int32, [4]>([1, 0, 0, 0])];
tensor<int32, [4]> k_31_end_0 = const()[name = tensor<string, []>("k_31_end_0"), val = tensor<int32, [4]>([2, 77, 1, 512])];
tensor<bool, [4]> k_31_end_mask_0 = const()[name = tensor<string, []>("k_31_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> k_31_squeeze_mask_0 = const()[name = tensor<string, []>("k_31_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> k_31_cast_fp16 = slice_by_index(begin = k_31_begin_0, end = k_31_end_0, end_mask = k_31_end_mask_0, squeeze_mask = k_31_squeeze_mask_0, x = var_617_cast_fp16)[name = tensor<string, []>("k_31_cast_fp16")];
tensor<int32, [4]> v_31_begin_0 = const()[name = tensor<string, []>("v_31_begin_0"), val = tensor<int32, [4]>([2, 0, 0, 0])];
tensor<int32, [4]> v_31_end_0 = const()[name = tensor<string, []>("v_31_end_0"), val = tensor<int32, [4]>([3, 77, 1, 512])];
tensor<bool, [4]> v_31_end_mask_0 = const()[name = tensor<string, []>("v_31_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> v_31_squeeze_mask_0 = const()[name = tensor<string, []>("v_31_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> v_31_cast_fp16 = slice_by_index(begin = v_31_begin_0, end = v_31_end_0, end_mask = v_31_end_mask_0, squeeze_mask = v_31_squeeze_mask_0, x = var_617_cast_fp16)[name = tensor<string, []>("v_31_cast_fp16")];
tensor<int32, [3]> var_626 = const()[name = tensor<string, []>("op_626"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_627_cast_fp16 = reshape(shape = var_626, x = q_31_cast_fp16)[name = tensor<string, []>("op_627_cast_fp16")];
tensor<int32, [3]> q_33_perm_0 = const()[name = tensor<string, []>("q_33_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_633 = const()[name = tensor<string, []>("op_633"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_634_cast_fp16 = reshape(shape = var_633, x = k_31_cast_fp16)[name = tensor<string, []>("op_634_cast_fp16")];
tensor<int32, [3]> k_33_perm_0 = const()[name = tensor<string, []>("k_33_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_640 = const()[name = tensor<string, []>("op_640"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_641_cast_fp16 = reshape(shape = var_640, x = v_31_cast_fp16)[name = tensor<string, []>("op_641_cast_fp16")];
tensor<int32, [3]> v_33_perm_0 = const()[name = tensor<string, []>("v_33_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [4]> var_645 = const()[name = tensor<string, []>("op_645"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> q_33_cast_fp16 = transpose(perm = q_33_perm_0, x = var_627_cast_fp16)[name = tensor<string, []>("transpose_48")];
tensor<fp16, [1, 8, 77, 64]> q_35_cast_fp16 = reshape(shape = var_645, x = q_33_cast_fp16)[name = tensor<string, []>("q_35_cast_fp16")];
tensor<int32, [4]> var_647 = const()[name = tensor<string, []>("op_647"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> k_33_cast_fp16 = transpose(perm = k_33_perm_0, x = var_634_cast_fp16)[name = tensor<string, []>("transpose_47")];
tensor<fp16, [1, 8, 77, 64]> k_35_cast_fp16 = reshape(shape = var_647, x = k_33_cast_fp16)[name = tensor<string, []>("k_35_cast_fp16")];
tensor<int32, [4]> var_649 = const()[name = tensor<string, []>("op_649"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> v_33_cast_fp16 = transpose(perm = v_33_perm_0, x = var_641_cast_fp16)[name = tensor<string, []>("transpose_46")];
tensor<fp16, [1, 8, 77, 64]> v_35_cast_fp16 = reshape(shape = var_649, x = v_33_cast_fp16)[name = tensor<string, []>("v_35_cast_fp16")];
tensor<fp16, []> mul_11_y_0_to_fp16 = const()[name = tensor<string, []>("mul_11_y_0_to_fp16"), val = tensor<fp16, []>(0x1p-3)];
tensor<fp16, [1, 8, 77, 64]> mul_11_cast_fp16 = mul(x = q_35_cast_fp16, y = mul_11_y_0_to_fp16)[name = tensor<string, []>("mul_11_cast_fp16")];
tensor<bool, []> matmul_5_transpose_y_0 = const()[name = tensor<string, []>("matmul_5_transpose_y_0"), val = tensor<bool, []>(true)];
tensor<bool, []> matmul_5_transpose_x_0 = const()[name = tensor<string, []>("matmul_5_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 77]> matmul_5_cast_fp16 = matmul(transpose_x = matmul_5_transpose_x_0, transpose_y = matmul_5_transpose_y_0, x = mul_11_cast_fp16, y = k_35_cast_fp16)[name = tensor<string, []>("matmul_5_cast_fp16")];
tensor<fp16, [1, 8, 77, 77]> add_5_cast_fp16 = add(x = matmul_5_cast_fp16, y = attn_mask_7_to_fp16)[name = tensor<string, []>("add_5_cast_fp16")];
tensor<int32, []> softmax_5_axis_0 = const()[name = tensor<string, []>("softmax_5_axis_0"), val = tensor<int32, []>(-1)];
tensor<fp16, [1, 8, 77, 77]> softmax_5_cast_fp16 = softmax(axis = softmax_5_axis_0, x = add_5_cast_fp16)[name = tensor<string, []>("softmax_5_cast_fp16")];
tensor<bool, []> attn_output_41_transpose_x_0 = const()[name = tensor<string, []>("attn_output_41_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<bool, []> attn_output_41_transpose_y_0 = const()[name = tensor<string, []>("attn_output_41_transpose_y_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 64]> attn_output_41_cast_fp16 = matmul(transpose_x = attn_output_41_transpose_x_0, transpose_y = attn_output_41_transpose_y_0, x = softmax_5_cast_fp16, y = v_35_cast_fp16)[name = tensor<string, []>("attn_output_41_cast_fp16")];
tensor<int32, [4]> var_652 = const()[name = tensor<string, []>("op_652"), val = tensor<int32, [4]>([2, 0, 1, 3])];
tensor<int32, [2]> var_657 = const()[name = tensor<string, []>("op_657"), val = tensor<int32, [2]>([77, 512])];
tensor<fp16, [77, 1, 8, 64]> var_653_cast_fp16 = transpose(perm = var_652, x = attn_output_41_cast_fp16)[name = tensor<string, []>("transpose_45")];
tensor<fp16, [77, 512]> attn_output_43_cast_fp16 = reshape(shape = var_657, x = var_653_cast_fp16)[name = tensor<string, []>("attn_output_43_cast_fp16")];
tensor<fp16, [512, 512]> model_transformer_resblocks_5_attn_out_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_5_attn_out_proj_weight_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(83790720)))];
tensor<fp16, [512]> model_transformer_resblocks_5_attn_out_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_5_attn_out_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(84315072)))];
tensor<fp16, [77, 512]> linear_21_cast_fp16 = linear(bias = model_transformer_resblocks_5_attn_out_proj_bias_to_fp16, weight = model_transformer_resblocks_5_attn_out_proj_weight_to_fp16, x = attn_output_43_cast_fp16)[name = tensor<string, []>("linear_21_cast_fp16")];
tensor<int32, [3]> var_661 = const()[name = tensor<string, []>("op_661"), val = tensor<int32, [3]>([77, 1, 512])];
tensor<fp16, [77, 1, 512]> attn_output_47_cast_fp16 = reshape(shape = var_661, x = linear_21_cast_fp16)[name = tensor<string, []>("attn_output_47_cast_fp16")];
tensor<int32, [3]> var_663_perm_0 = const()[name = tensor<string, []>("op_663_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1, 77, 512]> var_663_cast_fp16 = transpose(perm = var_663_perm_0, x = attn_output_47_cast_fp16)[name = tensor<string, []>("transpose_44")];
tensor<fp16, [1, 77, 512]> input_53_cast_fp16 = add(x = input_51_cast_fp16, y = var_663_cast_fp16)[name = tensor<string, []>("input_53_cast_fp16")];
tensor<int32, [1]> x_25_axes_0 = const()[name = tensor<string, []>("x_25_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_5_ln_2_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_5_ln_2_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(84316160)))];
tensor<fp16, [512]> model_transformer_resblocks_5_ln_2_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_5_ln_2_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(84317248)))];
tensor<fp16, [1, 77, 512]> x_25_cast_fp16 = layer_norm(axes = x_25_axes_0, beta = model_transformer_resblocks_5_ln_2_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_5_ln_2_weight_to_fp16, x = input_53_cast_fp16)[name = tensor<string, []>("x_25_cast_fp16")];
tensor<fp16, [2048, 512]> model_transformer_resblocks_5_mlp_c_fc_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_5_mlp_c_fc_weight_to_fp16"), val = tensor<fp16, [2048, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(84318336)))];
tensor<fp16, [2048]> model_transformer_resblocks_5_mlp_c_fc_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_5_mlp_c_fc_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(86415552)))];
tensor<fp16, [1, 77, 2048]> linear_22_cast_fp16 = linear(bias = model_transformer_resblocks_5_mlp_c_fc_bias_to_fp16, weight = model_transformer_resblocks_5_mlp_c_fc_weight_to_fp16, x = x_25_cast_fp16)[name = tensor<string, []>("linear_22_cast_fp16")];
tensor<string, []> input_59_mode_0 = const()[name = tensor<string, []>("input_59_mode_0"), val = tensor<string, []>("EXACT")];
tensor<fp16, [1, 77, 2048]> input_59_cast_fp16 = gelu(mode = input_59_mode_0, x = linear_22_cast_fp16)[name = tensor<string, []>("input_59_cast_fp16")];
tensor<fp16, [512, 2048]> model_transformer_resblocks_5_mlp_c_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_5_mlp_c_proj_weight_to_fp16"), val = tensor<fp16, [512, 2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(86419712)))];
tensor<fp16, [512]> model_transformer_resblocks_5_mlp_c_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_5_mlp_c_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(88516928)))];
tensor<fp16, [1, 77, 512]> linear_23_cast_fp16 = linear(bias = model_transformer_resblocks_5_mlp_c_proj_bias_to_fp16, weight = model_transformer_resblocks_5_mlp_c_proj_weight_to_fp16, x = input_59_cast_fp16)[name = tensor<string, []>("linear_23_cast_fp16")];
tensor<fp16, [1, 77, 512]> input_61_cast_fp16 = add(x = input_53_cast_fp16, y = linear_23_cast_fp16)[name = tensor<string, []>("input_61_cast_fp16")];
tensor<int32, [1]> x_27_axes_0 = const()[name = tensor<string, []>("x_27_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_6_ln_1_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_6_ln_1_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(88518016)))];
tensor<fp16, [512]> model_transformer_resblocks_6_ln_1_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_6_ln_1_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(88519104)))];
tensor<fp16, [1, 77, 512]> x_27_cast_fp16 = layer_norm(axes = x_27_axes_0, beta = model_transformer_resblocks_6_ln_1_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_6_ln_1_weight_to_fp16, x = input_61_cast_fp16)[name = tensor<string, []>("x_27_cast_fp16")];
tensor<int32, [3]> query_27_perm_0 = const()[name = tensor<string, []>("query_27_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1536, 512]> model_transformer_resblocks_6_attn_in_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_6_attn_in_proj_weight_to_fp16"), val = tensor<fp16, [1536, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(88520192)))];
tensor<fp16, [1536]> model_transformer_resblocks_6_attn_in_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_6_attn_in_proj_bias_to_fp16"), val = tensor<fp16, [1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(90093120)))];
tensor<fp16, [77, 1, 512]> query_27_cast_fp16 = transpose(perm = query_27_perm_0, x = x_27_cast_fp16)[name = tensor<string, []>("transpose_43")];
tensor<fp16, [77, 1, 1536]> linear_24_cast_fp16 = linear(bias = model_transformer_resblocks_6_attn_in_proj_bias_to_fp16, weight = model_transformer_resblocks_6_attn_in_proj_weight_to_fp16, x = query_27_cast_fp16)[name = tensor<string, []>("linear_24_cast_fp16")];
tensor<int32, [4]> concat_6 = const()[name = tensor<string, []>("concat_6"), val = tensor<int32, [4]>([77, 1, 3, 512])];
tensor<fp16, [77, 1, 3, 512]> var_716_cast_fp16 = reshape(shape = concat_6, x = linear_24_cast_fp16)[name = tensor<string, []>("op_716_cast_fp16")];
tensor<int32, [1]> var_717_axes_0 = const()[name = tensor<string, []>("op_717_axes_0"), val = tensor<int32, [1]>([0])];
tensor<fp16, [1, 77, 1, 3, 512]> var_717_cast_fp16 = expand_dims(axes = var_717_axes_0, x = var_716_cast_fp16)[name = tensor<string, []>("op_717_cast_fp16")];
tensor<int32, [5]> var_718_perm_0 = const()[name = tensor<string, []>("op_718_perm_0"), val = tensor<int32, [5]>([-2, 1, 2, 0, 4])];
tensor<int32, [1]> var_719_axes_0 = const()[name = tensor<string, []>("op_719_axes_0"), val = tensor<int32, [1]>([-2])];
tensor<fp16, [3, 77, 1, 1, 512]> var_718_cast_fp16 = transpose(perm = var_718_perm_0, x = var_717_cast_fp16)[name = tensor<string, []>("transpose_42")];
tensor<fp16, [3, 77, 1, 512]> var_719_cast_fp16 = squeeze(axes = var_719_axes_0, x = var_718_cast_fp16)[name = tensor<string, []>("op_719_cast_fp16")];
tensor<int32, [4]> q_37_begin_0 = const()[name = tensor<string, []>("q_37_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
tensor<int32, [4]> q_37_end_0 = const()[name = tensor<string, []>("q_37_end_0"), val = tensor<int32, [4]>([1, 77, 1, 512])];
tensor<bool, [4]> q_37_end_mask_0 = const()[name = tensor<string, []>("q_37_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> q_37_squeeze_mask_0 = const()[name = tensor<string, []>("q_37_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> q_37_cast_fp16 = slice_by_index(begin = q_37_begin_0, end = q_37_end_0, end_mask = q_37_end_mask_0, squeeze_mask = q_37_squeeze_mask_0, x = var_719_cast_fp16)[name = tensor<string, []>("q_37_cast_fp16")];
tensor<int32, [4]> k_37_begin_0 = const()[name = tensor<string, []>("k_37_begin_0"), val = tensor<int32, [4]>([1, 0, 0, 0])];
tensor<int32, [4]> k_37_end_0 = const()[name = tensor<string, []>("k_37_end_0"), val = tensor<int32, [4]>([2, 77, 1, 512])];
tensor<bool, [4]> k_37_end_mask_0 = const()[name = tensor<string, []>("k_37_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> k_37_squeeze_mask_0 = const()[name = tensor<string, []>("k_37_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> k_37_cast_fp16 = slice_by_index(begin = k_37_begin_0, end = k_37_end_0, end_mask = k_37_end_mask_0, squeeze_mask = k_37_squeeze_mask_0, x = var_719_cast_fp16)[name = tensor<string, []>("k_37_cast_fp16")];
tensor<int32, [4]> v_37_begin_0 = const()[name = tensor<string, []>("v_37_begin_0"), val = tensor<int32, [4]>([2, 0, 0, 0])];
tensor<int32, [4]> v_37_end_0 = const()[name = tensor<string, []>("v_37_end_0"), val = tensor<int32, [4]>([3, 77, 1, 512])];
tensor<bool, [4]> v_37_end_mask_0 = const()[name = tensor<string, []>("v_37_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> v_37_squeeze_mask_0 = const()[name = tensor<string, []>("v_37_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> v_37_cast_fp16 = slice_by_index(begin = v_37_begin_0, end = v_37_end_0, end_mask = v_37_end_mask_0, squeeze_mask = v_37_squeeze_mask_0, x = var_719_cast_fp16)[name = tensor<string, []>("v_37_cast_fp16")];
tensor<int32, [3]> var_728 = const()[name = tensor<string, []>("op_728"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_729_cast_fp16 = reshape(shape = var_728, x = q_37_cast_fp16)[name = tensor<string, []>("op_729_cast_fp16")];
tensor<int32, [3]> q_39_perm_0 = const()[name = tensor<string, []>("q_39_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_735 = const()[name = tensor<string, []>("op_735"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_736_cast_fp16 = reshape(shape = var_735, x = k_37_cast_fp16)[name = tensor<string, []>("op_736_cast_fp16")];
tensor<int32, [3]> k_39_perm_0 = const()[name = tensor<string, []>("k_39_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_742 = const()[name = tensor<string, []>("op_742"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_743_cast_fp16 = reshape(shape = var_742, x = v_37_cast_fp16)[name = tensor<string, []>("op_743_cast_fp16")];
tensor<int32, [3]> v_39_perm_0 = const()[name = tensor<string, []>("v_39_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [4]> var_747 = const()[name = tensor<string, []>("op_747"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> q_39_cast_fp16 = transpose(perm = q_39_perm_0, x = var_729_cast_fp16)[name = tensor<string, []>("transpose_41")];
tensor<fp16, [1, 8, 77, 64]> q_41_cast_fp16 = reshape(shape = var_747, x = q_39_cast_fp16)[name = tensor<string, []>("q_41_cast_fp16")];
tensor<int32, [4]> var_749 = const()[name = tensor<string, []>("op_749"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> k_39_cast_fp16 = transpose(perm = k_39_perm_0, x = var_736_cast_fp16)[name = tensor<string, []>("transpose_40")];
tensor<fp16, [1, 8, 77, 64]> k_41_cast_fp16 = reshape(shape = var_749, x = k_39_cast_fp16)[name = tensor<string, []>("k_41_cast_fp16")];
tensor<int32, [4]> var_751 = const()[name = tensor<string, []>("op_751"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> v_39_cast_fp16 = transpose(perm = v_39_perm_0, x = var_743_cast_fp16)[name = tensor<string, []>("transpose_39")];
tensor<fp16, [1, 8, 77, 64]> v_41_cast_fp16 = reshape(shape = var_751, x = v_39_cast_fp16)[name = tensor<string, []>("v_41_cast_fp16")];
tensor<fp16, []> mul_13_y_0_to_fp16 = const()[name = tensor<string, []>("mul_13_y_0_to_fp16"), val = tensor<fp16, []>(0x1p-3)];
tensor<fp16, [1, 8, 77, 64]> mul_13_cast_fp16 = mul(x = q_41_cast_fp16, y = mul_13_y_0_to_fp16)[name = tensor<string, []>("mul_13_cast_fp16")];
tensor<bool, []> matmul_6_transpose_y_0 = const()[name = tensor<string, []>("matmul_6_transpose_y_0"), val = tensor<bool, []>(true)];
tensor<bool, []> matmul_6_transpose_x_0 = const()[name = tensor<string, []>("matmul_6_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 77]> matmul_6_cast_fp16 = matmul(transpose_x = matmul_6_transpose_x_0, transpose_y = matmul_6_transpose_y_0, x = mul_13_cast_fp16, y = k_41_cast_fp16)[name = tensor<string, []>("matmul_6_cast_fp16")];
tensor<fp16, [1, 8, 77, 77]> add_6_cast_fp16 = add(x = matmul_6_cast_fp16, y = attn_mask_7_to_fp16)[name = tensor<string, []>("add_6_cast_fp16")];
tensor<int32, []> softmax_6_axis_0 = const()[name = tensor<string, []>("softmax_6_axis_0"), val = tensor<int32, []>(-1)];
tensor<fp16, [1, 8, 77, 77]> softmax_6_cast_fp16 = softmax(axis = softmax_6_axis_0, x = add_6_cast_fp16)[name = tensor<string, []>("softmax_6_cast_fp16")];
tensor<bool, []> attn_output_49_transpose_x_0 = const()[name = tensor<string, []>("attn_output_49_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<bool, []> attn_output_49_transpose_y_0 = const()[name = tensor<string, []>("attn_output_49_transpose_y_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 64]> attn_output_49_cast_fp16 = matmul(transpose_x = attn_output_49_transpose_x_0, transpose_y = attn_output_49_transpose_y_0, x = softmax_6_cast_fp16, y = v_41_cast_fp16)[name = tensor<string, []>("attn_output_49_cast_fp16")];
tensor<int32, [4]> var_754 = const()[name = tensor<string, []>("op_754"), val = tensor<int32, [4]>([2, 0, 1, 3])];
tensor<int32, [2]> var_759 = const()[name = tensor<string, []>("op_759"), val = tensor<int32, [2]>([77, 512])];
tensor<fp16, [77, 1, 8, 64]> var_755_cast_fp16 = transpose(perm = var_754, x = attn_output_49_cast_fp16)[name = tensor<string, []>("transpose_38")];
tensor<fp16, [77, 512]> attn_output_51_cast_fp16 = reshape(shape = var_759, x = var_755_cast_fp16)[name = tensor<string, []>("attn_output_51_cast_fp16")];
tensor<fp16, [512, 512]> model_transformer_resblocks_6_attn_out_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_6_attn_out_proj_weight_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(90096256)))];
tensor<fp16, [512]> model_transformer_resblocks_6_attn_out_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_6_attn_out_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(90620608)))];
tensor<fp16, [77, 512]> linear_25_cast_fp16 = linear(bias = model_transformer_resblocks_6_attn_out_proj_bias_to_fp16, weight = model_transformer_resblocks_6_attn_out_proj_weight_to_fp16, x = attn_output_51_cast_fp16)[name = tensor<string, []>("linear_25_cast_fp16")];
tensor<int32, [3]> var_763 = const()[name = tensor<string, []>("op_763"), val = tensor<int32, [3]>([77, 1, 512])];
tensor<fp16, [77, 1, 512]> attn_output_55_cast_fp16 = reshape(shape = var_763, x = linear_25_cast_fp16)[name = tensor<string, []>("attn_output_55_cast_fp16")];
tensor<int32, [3]> var_765_perm_0 = const()[name = tensor<string, []>("op_765_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1, 77, 512]> var_765_cast_fp16 = transpose(perm = var_765_perm_0, x = attn_output_55_cast_fp16)[name = tensor<string, []>("transpose_37")];
tensor<fp16, [1, 77, 512]> input_63_cast_fp16 = add(x = input_61_cast_fp16, y = var_765_cast_fp16)[name = tensor<string, []>("input_63_cast_fp16")];
tensor<int32, [1]> x_29_axes_0 = const()[name = tensor<string, []>("x_29_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_6_ln_2_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_6_ln_2_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(90621696)))];
tensor<fp16, [512]> model_transformer_resblocks_6_ln_2_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_6_ln_2_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(90622784)))];
tensor<fp16, [1, 77, 512]> x_29_cast_fp16 = layer_norm(axes = x_29_axes_0, beta = model_transformer_resblocks_6_ln_2_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_6_ln_2_weight_to_fp16, x = input_63_cast_fp16)[name = tensor<string, []>("x_29_cast_fp16")];
tensor<fp16, [2048, 512]> model_transformer_resblocks_6_mlp_c_fc_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_6_mlp_c_fc_weight_to_fp16"), val = tensor<fp16, [2048, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(90623872)))];
tensor<fp16, [2048]> model_transformer_resblocks_6_mlp_c_fc_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_6_mlp_c_fc_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(92721088)))];
tensor<fp16, [1, 77, 2048]> linear_26_cast_fp16 = linear(bias = model_transformer_resblocks_6_mlp_c_fc_bias_to_fp16, weight = model_transformer_resblocks_6_mlp_c_fc_weight_to_fp16, x = x_29_cast_fp16)[name = tensor<string, []>("linear_26_cast_fp16")];
tensor<string, []> input_69_mode_0 = const()[name = tensor<string, []>("input_69_mode_0"), val = tensor<string, []>("EXACT")];
tensor<fp16, [1, 77, 2048]> input_69_cast_fp16 = gelu(mode = input_69_mode_0, x = linear_26_cast_fp16)[name = tensor<string, []>("input_69_cast_fp16")];
tensor<fp16, [512, 2048]> model_transformer_resblocks_6_mlp_c_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_6_mlp_c_proj_weight_to_fp16"), val = tensor<fp16, [512, 2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(92725248)))];
tensor<fp16, [512]> model_transformer_resblocks_6_mlp_c_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_6_mlp_c_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(94822464)))];
tensor<fp16, [1, 77, 512]> linear_27_cast_fp16 = linear(bias = model_transformer_resblocks_6_mlp_c_proj_bias_to_fp16, weight = model_transformer_resblocks_6_mlp_c_proj_weight_to_fp16, x = input_69_cast_fp16)[name = tensor<string, []>("linear_27_cast_fp16")];
tensor<fp16, [1, 77, 512]> input_71_cast_fp16 = add(x = input_63_cast_fp16, y = linear_27_cast_fp16)[name = tensor<string, []>("input_71_cast_fp16")];
tensor<int32, [1]> x_31_axes_0 = const()[name = tensor<string, []>("x_31_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_7_ln_1_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_7_ln_1_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(94823552)))];
tensor<fp16, [512]> model_transformer_resblocks_7_ln_1_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_7_ln_1_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(94824640)))];
tensor<fp16, [1, 77, 512]> x_31_cast_fp16 = layer_norm(axes = x_31_axes_0, beta = model_transformer_resblocks_7_ln_1_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_7_ln_1_weight_to_fp16, x = input_71_cast_fp16)[name = tensor<string, []>("x_31_cast_fp16")];
tensor<int32, [3]> query_31_perm_0 = const()[name = tensor<string, []>("query_31_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1536, 512]> model_transformer_resblocks_7_attn_in_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_7_attn_in_proj_weight_to_fp16"), val = tensor<fp16, [1536, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(94825728)))];
tensor<fp16, [1536]> model_transformer_resblocks_7_attn_in_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_7_attn_in_proj_bias_to_fp16"), val = tensor<fp16, [1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(96398656)))];
tensor<fp16, [77, 1, 512]> query_31_cast_fp16 = transpose(perm = query_31_perm_0, x = x_31_cast_fp16)[name = tensor<string, []>("transpose_36")];
tensor<fp16, [77, 1, 1536]> linear_28_cast_fp16 = linear(bias = model_transformer_resblocks_7_attn_in_proj_bias_to_fp16, weight = model_transformer_resblocks_7_attn_in_proj_weight_to_fp16, x = query_31_cast_fp16)[name = tensor<string, []>("linear_28_cast_fp16")];
tensor<int32, [4]> concat_7 = const()[name = tensor<string, []>("concat_7"), val = tensor<int32, [4]>([77, 1, 3, 512])];
tensor<fp16, [77, 1, 3, 512]> var_818_cast_fp16 = reshape(shape = concat_7, x = linear_28_cast_fp16)[name = tensor<string, []>("op_818_cast_fp16")];
tensor<int32, [1]> var_819_axes_0 = const()[name = tensor<string, []>("op_819_axes_0"), val = tensor<int32, [1]>([0])];
tensor<fp16, [1, 77, 1, 3, 512]> var_819_cast_fp16 = expand_dims(axes = var_819_axes_0, x = var_818_cast_fp16)[name = tensor<string, []>("op_819_cast_fp16")];
tensor<int32, [5]> var_820_perm_0 = const()[name = tensor<string, []>("op_820_perm_0"), val = tensor<int32, [5]>([-2, 1, 2, 0, 4])];
tensor<int32, [1]> var_821_axes_0 = const()[name = tensor<string, []>("op_821_axes_0"), val = tensor<int32, [1]>([-2])];
tensor<fp16, [3, 77, 1, 1, 512]> var_820_cast_fp16 = transpose(perm = var_820_perm_0, x = var_819_cast_fp16)[name = tensor<string, []>("transpose_35")];
tensor<fp16, [3, 77, 1, 512]> var_821_cast_fp16 = squeeze(axes = var_821_axes_0, x = var_820_cast_fp16)[name = tensor<string, []>("op_821_cast_fp16")];
tensor<int32, [4]> q_43_begin_0 = const()[name = tensor<string, []>("q_43_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
tensor<int32, [4]> q_43_end_0 = const()[name = tensor<string, []>("q_43_end_0"), val = tensor<int32, [4]>([1, 77, 1, 512])];
tensor<bool, [4]> q_43_end_mask_0 = const()[name = tensor<string, []>("q_43_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> q_43_squeeze_mask_0 = const()[name = tensor<string, []>("q_43_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> q_43_cast_fp16 = slice_by_index(begin = q_43_begin_0, end = q_43_end_0, end_mask = q_43_end_mask_0, squeeze_mask = q_43_squeeze_mask_0, x = var_821_cast_fp16)[name = tensor<string, []>("q_43_cast_fp16")];
tensor<int32, [4]> k_43_begin_0 = const()[name = tensor<string, []>("k_43_begin_0"), val = tensor<int32, [4]>([1, 0, 0, 0])];
tensor<int32, [4]> k_43_end_0 = const()[name = tensor<string, []>("k_43_end_0"), val = tensor<int32, [4]>([2, 77, 1, 512])];
tensor<bool, [4]> k_43_end_mask_0 = const()[name = tensor<string, []>("k_43_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> k_43_squeeze_mask_0 = const()[name = tensor<string, []>("k_43_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> k_43_cast_fp16 = slice_by_index(begin = k_43_begin_0, end = k_43_end_0, end_mask = k_43_end_mask_0, squeeze_mask = k_43_squeeze_mask_0, x = var_821_cast_fp16)[name = tensor<string, []>("k_43_cast_fp16")];
tensor<int32, [4]> v_43_begin_0 = const()[name = tensor<string, []>("v_43_begin_0"), val = tensor<int32, [4]>([2, 0, 0, 0])];
tensor<int32, [4]> v_43_end_0 = const()[name = tensor<string, []>("v_43_end_0"), val = tensor<int32, [4]>([3, 77, 1, 512])];
tensor<bool, [4]> v_43_end_mask_0 = const()[name = tensor<string, []>("v_43_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> v_43_squeeze_mask_0 = const()[name = tensor<string, []>("v_43_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> v_43_cast_fp16 = slice_by_index(begin = v_43_begin_0, end = v_43_end_0, end_mask = v_43_end_mask_0, squeeze_mask = v_43_squeeze_mask_0, x = var_821_cast_fp16)[name = tensor<string, []>("v_43_cast_fp16")];
tensor<int32, [3]> var_830 = const()[name = tensor<string, []>("op_830"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_831_cast_fp16 = reshape(shape = var_830, x = q_43_cast_fp16)[name = tensor<string, []>("op_831_cast_fp16")];
tensor<int32, [3]> q_45_perm_0 = const()[name = tensor<string, []>("q_45_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_837 = const()[name = tensor<string, []>("op_837"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_838_cast_fp16 = reshape(shape = var_837, x = k_43_cast_fp16)[name = tensor<string, []>("op_838_cast_fp16")];
tensor<int32, [3]> k_45_perm_0 = const()[name = tensor<string, []>("k_45_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_844 = const()[name = tensor<string, []>("op_844"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_845_cast_fp16 = reshape(shape = var_844, x = v_43_cast_fp16)[name = tensor<string, []>("op_845_cast_fp16")];
tensor<int32, [3]> v_45_perm_0 = const()[name = tensor<string, []>("v_45_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [4]> var_849 = const()[name = tensor<string, []>("op_849"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> q_45_cast_fp16 = transpose(perm = q_45_perm_0, x = var_831_cast_fp16)[name = tensor<string, []>("transpose_34")];
tensor<fp16, [1, 8, 77, 64]> q_47_cast_fp16 = reshape(shape = var_849, x = q_45_cast_fp16)[name = tensor<string, []>("q_47_cast_fp16")];
tensor<int32, [4]> var_851 = const()[name = tensor<string, []>("op_851"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> k_45_cast_fp16 = transpose(perm = k_45_perm_0, x = var_838_cast_fp16)[name = tensor<string, []>("transpose_33")];
tensor<fp16, [1, 8, 77, 64]> k_47_cast_fp16 = reshape(shape = var_851, x = k_45_cast_fp16)[name = tensor<string, []>("k_47_cast_fp16")];
tensor<int32, [4]> var_853 = const()[name = tensor<string, []>("op_853"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> v_45_cast_fp16 = transpose(perm = v_45_perm_0, x = var_845_cast_fp16)[name = tensor<string, []>("transpose_32")];
tensor<fp16, [1, 8, 77, 64]> v_47_cast_fp16 = reshape(shape = var_853, x = v_45_cast_fp16)[name = tensor<string, []>("v_47_cast_fp16")];
tensor<fp16, []> mul_15_y_0_to_fp16 = const()[name = tensor<string, []>("mul_15_y_0_to_fp16"), val = tensor<fp16, []>(0x1p-3)];
tensor<fp16, [1, 8, 77, 64]> mul_15_cast_fp16 = mul(x = q_47_cast_fp16, y = mul_15_y_0_to_fp16)[name = tensor<string, []>("mul_15_cast_fp16")];
tensor<bool, []> matmul_7_transpose_y_0 = const()[name = tensor<string, []>("matmul_7_transpose_y_0"), val = tensor<bool, []>(true)];
tensor<bool, []> matmul_7_transpose_x_0 = const()[name = tensor<string, []>("matmul_7_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 77]> matmul_7_cast_fp16 = matmul(transpose_x = matmul_7_transpose_x_0, transpose_y = matmul_7_transpose_y_0, x = mul_15_cast_fp16, y = k_47_cast_fp16)[name = tensor<string, []>("matmul_7_cast_fp16")];
tensor<fp16, [1, 8, 77, 77]> add_7_cast_fp16 = add(x = matmul_7_cast_fp16, y = attn_mask_7_to_fp16)[name = tensor<string, []>("add_7_cast_fp16")];
tensor<int32, []> softmax_7_axis_0 = const()[name = tensor<string, []>("softmax_7_axis_0"), val = tensor<int32, []>(-1)];
tensor<fp16, [1, 8, 77, 77]> softmax_7_cast_fp16 = softmax(axis = softmax_7_axis_0, x = add_7_cast_fp16)[name = tensor<string, []>("softmax_7_cast_fp16")];
tensor<bool, []> attn_output_57_transpose_x_0 = const()[name = tensor<string, []>("attn_output_57_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<bool, []> attn_output_57_transpose_y_0 = const()[name = tensor<string, []>("attn_output_57_transpose_y_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 64]> attn_output_57_cast_fp16 = matmul(transpose_x = attn_output_57_transpose_x_0, transpose_y = attn_output_57_transpose_y_0, x = softmax_7_cast_fp16, y = v_47_cast_fp16)[name = tensor<string, []>("attn_output_57_cast_fp16")];
tensor<int32, [4]> var_856 = const()[name = tensor<string, []>("op_856"), val = tensor<int32, [4]>([2, 0, 1, 3])];
tensor<int32, [2]> var_861 = const()[name = tensor<string, []>("op_861"), val = tensor<int32, [2]>([77, 512])];
tensor<fp16, [77, 1, 8, 64]> var_857_cast_fp16 = transpose(perm = var_856, x = attn_output_57_cast_fp16)[name = tensor<string, []>("transpose_31")];
tensor<fp16, [77, 512]> attn_output_59_cast_fp16 = reshape(shape = var_861, x = var_857_cast_fp16)[name = tensor<string, []>("attn_output_59_cast_fp16")];
tensor<fp16, [512, 512]> model_transformer_resblocks_7_attn_out_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_7_attn_out_proj_weight_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(96401792)))];
tensor<fp16, [512]> model_transformer_resblocks_7_attn_out_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_7_attn_out_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(96926144)))];
tensor<fp16, [77, 512]> linear_29_cast_fp16 = linear(bias = model_transformer_resblocks_7_attn_out_proj_bias_to_fp16, weight = model_transformer_resblocks_7_attn_out_proj_weight_to_fp16, x = attn_output_59_cast_fp16)[name = tensor<string, []>("linear_29_cast_fp16")];
tensor<int32, [3]> var_865 = const()[name = tensor<string, []>("op_865"), val = tensor<int32, [3]>([77, 1, 512])];
tensor<fp16, [77, 1, 512]> attn_output_63_cast_fp16 = reshape(shape = var_865, x = linear_29_cast_fp16)[name = tensor<string, []>("attn_output_63_cast_fp16")];
tensor<int32, [3]> var_867_perm_0 = const()[name = tensor<string, []>("op_867_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1, 77, 512]> var_867_cast_fp16 = transpose(perm = var_867_perm_0, x = attn_output_63_cast_fp16)[name = tensor<string, []>("transpose_30")];
tensor<fp16, [1, 77, 512]> input_73_cast_fp16 = add(x = input_71_cast_fp16, y = var_867_cast_fp16)[name = tensor<string, []>("input_73_cast_fp16")];
tensor<int32, [1]> x_33_axes_0 = const()[name = tensor<string, []>("x_33_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_7_ln_2_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_7_ln_2_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(96927232)))];
tensor<fp16, [512]> model_transformer_resblocks_7_ln_2_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_7_ln_2_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(96928320)))];
tensor<fp16, [1, 77, 512]> x_33_cast_fp16 = layer_norm(axes = x_33_axes_0, beta = model_transformer_resblocks_7_ln_2_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_7_ln_2_weight_to_fp16, x = input_73_cast_fp16)[name = tensor<string, []>("x_33_cast_fp16")];
tensor<fp16, [2048, 512]> model_transformer_resblocks_7_mlp_c_fc_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_7_mlp_c_fc_weight_to_fp16"), val = tensor<fp16, [2048, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(96929408)))];
tensor<fp16, [2048]> model_transformer_resblocks_7_mlp_c_fc_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_7_mlp_c_fc_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(99026624)))];
tensor<fp16, [1, 77, 2048]> linear_30_cast_fp16 = linear(bias = model_transformer_resblocks_7_mlp_c_fc_bias_to_fp16, weight = model_transformer_resblocks_7_mlp_c_fc_weight_to_fp16, x = x_33_cast_fp16)[name = tensor<string, []>("linear_30_cast_fp16")];
tensor<string, []> input_79_mode_0 = const()[name = tensor<string, []>("input_79_mode_0"), val = tensor<string, []>("EXACT")];
tensor<fp16, [1, 77, 2048]> input_79_cast_fp16 = gelu(mode = input_79_mode_0, x = linear_30_cast_fp16)[name = tensor<string, []>("input_79_cast_fp16")];
tensor<fp16, [512, 2048]> model_transformer_resblocks_7_mlp_c_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_7_mlp_c_proj_weight_to_fp16"), val = tensor<fp16, [512, 2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(99030784)))];
tensor<fp16, [512]> model_transformer_resblocks_7_mlp_c_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_7_mlp_c_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(101128000)))];
tensor<fp16, [1, 77, 512]> linear_31_cast_fp16 = linear(bias = model_transformer_resblocks_7_mlp_c_proj_bias_to_fp16, weight = model_transformer_resblocks_7_mlp_c_proj_weight_to_fp16, x = input_79_cast_fp16)[name = tensor<string, []>("linear_31_cast_fp16")];
tensor<fp16, [1, 77, 512]> input_81_cast_fp16 = add(x = input_73_cast_fp16, y = linear_31_cast_fp16)[name = tensor<string, []>("input_81_cast_fp16")];
tensor<int32, [1]> x_35_axes_0 = const()[name = tensor<string, []>("x_35_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_8_ln_1_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_8_ln_1_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(101129088)))];
tensor<fp16, [512]> model_transformer_resblocks_8_ln_1_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_8_ln_1_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(101130176)))];
tensor<fp16, [1, 77, 512]> x_35_cast_fp16 = layer_norm(axes = x_35_axes_0, beta = model_transformer_resblocks_8_ln_1_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_8_ln_1_weight_to_fp16, x = input_81_cast_fp16)[name = tensor<string, []>("x_35_cast_fp16")];
tensor<int32, [3]> query_35_perm_0 = const()[name = tensor<string, []>("query_35_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1536, 512]> model_transformer_resblocks_8_attn_in_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_8_attn_in_proj_weight_to_fp16"), val = tensor<fp16, [1536, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(101131264)))];
tensor<fp16, [1536]> model_transformer_resblocks_8_attn_in_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_8_attn_in_proj_bias_to_fp16"), val = tensor<fp16, [1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(102704192)))];
tensor<fp16, [77, 1, 512]> query_35_cast_fp16 = transpose(perm = query_35_perm_0, x = x_35_cast_fp16)[name = tensor<string, []>("transpose_29")];
tensor<fp16, [77, 1, 1536]> linear_32_cast_fp16 = linear(bias = model_transformer_resblocks_8_attn_in_proj_bias_to_fp16, weight = model_transformer_resblocks_8_attn_in_proj_weight_to_fp16, x = query_35_cast_fp16)[name = tensor<string, []>("linear_32_cast_fp16")];
tensor<int32, [4]> concat_8 = const()[name = tensor<string, []>("concat_8"), val = tensor<int32, [4]>([77, 1, 3, 512])];
tensor<fp16, [77, 1, 3, 512]> var_920_cast_fp16 = reshape(shape = concat_8, x = linear_32_cast_fp16)[name = tensor<string, []>("op_920_cast_fp16")];
tensor<int32, [1]> var_921_axes_0 = const()[name = tensor<string, []>("op_921_axes_0"), val = tensor<int32, [1]>([0])];
tensor<fp16, [1, 77, 1, 3, 512]> var_921_cast_fp16 = expand_dims(axes = var_921_axes_0, x = var_920_cast_fp16)[name = tensor<string, []>("op_921_cast_fp16")];
tensor<int32, [5]> var_922_perm_0 = const()[name = tensor<string, []>("op_922_perm_0"), val = tensor<int32, [5]>([-2, 1, 2, 0, 4])];
tensor<int32, [1]> var_923_axes_0 = const()[name = tensor<string, []>("op_923_axes_0"), val = tensor<int32, [1]>([-2])];
tensor<fp16, [3, 77, 1, 1, 512]> var_922_cast_fp16 = transpose(perm = var_922_perm_0, x = var_921_cast_fp16)[name = tensor<string, []>("transpose_28")];
tensor<fp16, [3, 77, 1, 512]> var_923_cast_fp16 = squeeze(axes = var_923_axes_0, x = var_922_cast_fp16)[name = tensor<string, []>("op_923_cast_fp16")];
tensor<int32, [4]> q_49_begin_0 = const()[name = tensor<string, []>("q_49_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
tensor<int32, [4]> q_49_end_0 = const()[name = tensor<string, []>("q_49_end_0"), val = tensor<int32, [4]>([1, 77, 1, 512])];
tensor<bool, [4]> q_49_end_mask_0 = const()[name = tensor<string, []>("q_49_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> q_49_squeeze_mask_0 = const()[name = tensor<string, []>("q_49_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> q_49_cast_fp16 = slice_by_index(begin = q_49_begin_0, end = q_49_end_0, end_mask = q_49_end_mask_0, squeeze_mask = q_49_squeeze_mask_0, x = var_923_cast_fp16)[name = tensor<string, []>("q_49_cast_fp16")];
tensor<int32, [4]> k_49_begin_0 = const()[name = tensor<string, []>("k_49_begin_0"), val = tensor<int32, [4]>([1, 0, 0, 0])];
tensor<int32, [4]> k_49_end_0 = const()[name = tensor<string, []>("k_49_end_0"), val = tensor<int32, [4]>([2, 77, 1, 512])];
tensor<bool, [4]> k_49_end_mask_0 = const()[name = tensor<string, []>("k_49_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> k_49_squeeze_mask_0 = const()[name = tensor<string, []>("k_49_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> k_49_cast_fp16 = slice_by_index(begin = k_49_begin_0, end = k_49_end_0, end_mask = k_49_end_mask_0, squeeze_mask = k_49_squeeze_mask_0, x = var_923_cast_fp16)[name = tensor<string, []>("k_49_cast_fp16")];
tensor<int32, [4]> v_49_begin_0 = const()[name = tensor<string, []>("v_49_begin_0"), val = tensor<int32, [4]>([2, 0, 0, 0])];
tensor<int32, [4]> v_49_end_0 = const()[name = tensor<string, []>("v_49_end_0"), val = tensor<int32, [4]>([3, 77, 1, 512])];
tensor<bool, [4]> v_49_end_mask_0 = const()[name = tensor<string, []>("v_49_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> v_49_squeeze_mask_0 = const()[name = tensor<string, []>("v_49_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> v_49_cast_fp16 = slice_by_index(begin = v_49_begin_0, end = v_49_end_0, end_mask = v_49_end_mask_0, squeeze_mask = v_49_squeeze_mask_0, x = var_923_cast_fp16)[name = tensor<string, []>("v_49_cast_fp16")];
tensor<int32, [3]> var_932 = const()[name = tensor<string, []>("op_932"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_933_cast_fp16 = reshape(shape = var_932, x = q_49_cast_fp16)[name = tensor<string, []>("op_933_cast_fp16")];
tensor<int32, [3]> q_51_perm_0 = const()[name = tensor<string, []>("q_51_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_939 = const()[name = tensor<string, []>("op_939"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_940_cast_fp16 = reshape(shape = var_939, x = k_49_cast_fp16)[name = tensor<string, []>("op_940_cast_fp16")];
tensor<int32, [3]> k_51_perm_0 = const()[name = tensor<string, []>("k_51_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_946 = const()[name = tensor<string, []>("op_946"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_947_cast_fp16 = reshape(shape = var_946, x = v_49_cast_fp16)[name = tensor<string, []>("op_947_cast_fp16")];
tensor<int32, [3]> v_51_perm_0 = const()[name = tensor<string, []>("v_51_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [4]> var_951 = const()[name = tensor<string, []>("op_951"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> q_51_cast_fp16 = transpose(perm = q_51_perm_0, x = var_933_cast_fp16)[name = tensor<string, []>("transpose_27")];
tensor<fp16, [1, 8, 77, 64]> q_53_cast_fp16 = reshape(shape = var_951, x = q_51_cast_fp16)[name = tensor<string, []>("q_53_cast_fp16")];
tensor<int32, [4]> var_953 = const()[name = tensor<string, []>("op_953"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> k_51_cast_fp16 = transpose(perm = k_51_perm_0, x = var_940_cast_fp16)[name = tensor<string, []>("transpose_26")];
tensor<fp16, [1, 8, 77, 64]> k_53_cast_fp16 = reshape(shape = var_953, x = k_51_cast_fp16)[name = tensor<string, []>("k_53_cast_fp16")];
tensor<int32, [4]> var_955 = const()[name = tensor<string, []>("op_955"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> v_51_cast_fp16 = transpose(perm = v_51_perm_0, x = var_947_cast_fp16)[name = tensor<string, []>("transpose_25")];
tensor<fp16, [1, 8, 77, 64]> v_53_cast_fp16 = reshape(shape = var_955, x = v_51_cast_fp16)[name = tensor<string, []>("v_53_cast_fp16")];
tensor<fp16, []> mul_17_y_0_to_fp16 = const()[name = tensor<string, []>("mul_17_y_0_to_fp16"), val = tensor<fp16, []>(0x1p-3)];
tensor<fp16, [1, 8, 77, 64]> mul_17_cast_fp16 = mul(x = q_53_cast_fp16, y = mul_17_y_0_to_fp16)[name = tensor<string, []>("mul_17_cast_fp16")];
tensor<bool, []> matmul_8_transpose_y_0 = const()[name = tensor<string, []>("matmul_8_transpose_y_0"), val = tensor<bool, []>(true)];
tensor<bool, []> matmul_8_transpose_x_0 = const()[name = tensor<string, []>("matmul_8_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 77]> matmul_8_cast_fp16 = matmul(transpose_x = matmul_8_transpose_x_0, transpose_y = matmul_8_transpose_y_0, x = mul_17_cast_fp16, y = k_53_cast_fp16)[name = tensor<string, []>("matmul_8_cast_fp16")];
tensor<fp16, [1, 8, 77, 77]> add_8_cast_fp16 = add(x = matmul_8_cast_fp16, y = attn_mask_7_to_fp16)[name = tensor<string, []>("add_8_cast_fp16")];
tensor<int32, []> softmax_8_axis_0 = const()[name = tensor<string, []>("softmax_8_axis_0"), val = tensor<int32, []>(-1)];
tensor<fp16, [1, 8, 77, 77]> softmax_8_cast_fp16 = softmax(axis = softmax_8_axis_0, x = add_8_cast_fp16)[name = tensor<string, []>("softmax_8_cast_fp16")];
tensor<bool, []> attn_output_65_transpose_x_0 = const()[name = tensor<string, []>("attn_output_65_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<bool, []> attn_output_65_transpose_y_0 = const()[name = tensor<string, []>("attn_output_65_transpose_y_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 64]> attn_output_65_cast_fp16 = matmul(transpose_x = attn_output_65_transpose_x_0, transpose_y = attn_output_65_transpose_y_0, x = softmax_8_cast_fp16, y = v_53_cast_fp16)[name = tensor<string, []>("attn_output_65_cast_fp16")];
tensor<int32, [4]> var_958 = const()[name = tensor<string, []>("op_958"), val = tensor<int32, [4]>([2, 0, 1, 3])];
tensor<int32, [2]> var_963 = const()[name = tensor<string, []>("op_963"), val = tensor<int32, [2]>([77, 512])];
tensor<fp16, [77, 1, 8, 64]> var_959_cast_fp16 = transpose(perm = var_958, x = attn_output_65_cast_fp16)[name = tensor<string, []>("transpose_24")];
tensor<fp16, [77, 512]> attn_output_67_cast_fp16 = reshape(shape = var_963, x = var_959_cast_fp16)[name = tensor<string, []>("attn_output_67_cast_fp16")];
tensor<fp16, [512, 512]> model_transformer_resblocks_8_attn_out_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_8_attn_out_proj_weight_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(102707328)))];
tensor<fp16, [512]> model_transformer_resblocks_8_attn_out_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_8_attn_out_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(103231680)))];
tensor<fp16, [77, 512]> linear_33_cast_fp16 = linear(bias = model_transformer_resblocks_8_attn_out_proj_bias_to_fp16, weight = model_transformer_resblocks_8_attn_out_proj_weight_to_fp16, x = attn_output_67_cast_fp16)[name = tensor<string, []>("linear_33_cast_fp16")];
tensor<int32, [3]> var_967 = const()[name = tensor<string, []>("op_967"), val = tensor<int32, [3]>([77, 1, 512])];
tensor<fp16, [77, 1, 512]> attn_output_71_cast_fp16 = reshape(shape = var_967, x = linear_33_cast_fp16)[name = tensor<string, []>("attn_output_71_cast_fp16")];
tensor<int32, [3]> var_969_perm_0 = const()[name = tensor<string, []>("op_969_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1, 77, 512]> var_969_cast_fp16 = transpose(perm = var_969_perm_0, x = attn_output_71_cast_fp16)[name = tensor<string, []>("transpose_23")];
tensor<fp16, [1, 77, 512]> input_83_cast_fp16 = add(x = input_81_cast_fp16, y = var_969_cast_fp16)[name = tensor<string, []>("input_83_cast_fp16")];
tensor<int32, [1]> x_37_axes_0 = const()[name = tensor<string, []>("x_37_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_8_ln_2_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_8_ln_2_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(103232768)))];
tensor<fp16, [512]> model_transformer_resblocks_8_ln_2_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_8_ln_2_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(103233856)))];
tensor<fp16, [1, 77, 512]> x_37_cast_fp16 = layer_norm(axes = x_37_axes_0, beta = model_transformer_resblocks_8_ln_2_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_8_ln_2_weight_to_fp16, x = input_83_cast_fp16)[name = tensor<string, []>("x_37_cast_fp16")];
tensor<fp16, [2048, 512]> model_transformer_resblocks_8_mlp_c_fc_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_8_mlp_c_fc_weight_to_fp16"), val = tensor<fp16, [2048, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(103234944)))];
tensor<fp16, [2048]> model_transformer_resblocks_8_mlp_c_fc_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_8_mlp_c_fc_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(105332160)))];
tensor<fp16, [1, 77, 2048]> linear_34_cast_fp16 = linear(bias = model_transformer_resblocks_8_mlp_c_fc_bias_to_fp16, weight = model_transformer_resblocks_8_mlp_c_fc_weight_to_fp16, x = x_37_cast_fp16)[name = tensor<string, []>("linear_34_cast_fp16")];
tensor<string, []> input_89_mode_0 = const()[name = tensor<string, []>("input_89_mode_0"), val = tensor<string, []>("EXACT")];
tensor<fp16, [1, 77, 2048]> input_89_cast_fp16 = gelu(mode = input_89_mode_0, x = linear_34_cast_fp16)[name = tensor<string, []>("input_89_cast_fp16")];
tensor<fp16, [512, 2048]> model_transformer_resblocks_8_mlp_c_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_8_mlp_c_proj_weight_to_fp16"), val = tensor<fp16, [512, 2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(105336320)))];
tensor<fp16, [512]> model_transformer_resblocks_8_mlp_c_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_8_mlp_c_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(107433536)))];
tensor<fp16, [1, 77, 512]> linear_35_cast_fp16 = linear(bias = model_transformer_resblocks_8_mlp_c_proj_bias_to_fp16, weight = model_transformer_resblocks_8_mlp_c_proj_weight_to_fp16, x = input_89_cast_fp16)[name = tensor<string, []>("linear_35_cast_fp16")];
tensor<fp16, [1, 77, 512]> input_91_cast_fp16 = add(x = input_83_cast_fp16, y = linear_35_cast_fp16)[name = tensor<string, []>("input_91_cast_fp16")];
tensor<int32, [1]> x_39_axes_0 = const()[name = tensor<string, []>("x_39_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_9_ln_1_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_9_ln_1_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(107434624)))];
tensor<fp16, [512]> model_transformer_resblocks_9_ln_1_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_9_ln_1_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(107435712)))];
tensor<fp16, [1, 77, 512]> x_39_cast_fp16 = layer_norm(axes = x_39_axes_0, beta = model_transformer_resblocks_9_ln_1_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_9_ln_1_weight_to_fp16, x = input_91_cast_fp16)[name = tensor<string, []>("x_39_cast_fp16")];
tensor<int32, [3]> query_39_perm_0 = const()[name = tensor<string, []>("query_39_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1536, 512]> model_transformer_resblocks_9_attn_in_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_9_attn_in_proj_weight_to_fp16"), val = tensor<fp16, [1536, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(107436800)))];
tensor<fp16, [1536]> model_transformer_resblocks_9_attn_in_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_9_attn_in_proj_bias_to_fp16"), val = tensor<fp16, [1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(109009728)))];
tensor<fp16, [77, 1, 512]> query_39_cast_fp16 = transpose(perm = query_39_perm_0, x = x_39_cast_fp16)[name = tensor<string, []>("transpose_22")];
tensor<fp16, [77, 1, 1536]> linear_36_cast_fp16 = linear(bias = model_transformer_resblocks_9_attn_in_proj_bias_to_fp16, weight = model_transformer_resblocks_9_attn_in_proj_weight_to_fp16, x = query_39_cast_fp16)[name = tensor<string, []>("linear_36_cast_fp16")];
tensor<int32, [4]> concat_9 = const()[name = tensor<string, []>("concat_9"), val = tensor<int32, [4]>([77, 1, 3, 512])];
tensor<fp16, [77, 1, 3, 512]> var_1022_cast_fp16 = reshape(shape = concat_9, x = linear_36_cast_fp16)[name = tensor<string, []>("op_1022_cast_fp16")];
tensor<int32, [1]> var_1023_axes_0 = const()[name = tensor<string, []>("op_1023_axes_0"), val = tensor<int32, [1]>([0])];
tensor<fp16, [1, 77, 1, 3, 512]> var_1023_cast_fp16 = expand_dims(axes = var_1023_axes_0, x = var_1022_cast_fp16)[name = tensor<string, []>("op_1023_cast_fp16")];
tensor<int32, [5]> var_1024_perm_0 = const()[name = tensor<string, []>("op_1024_perm_0"), val = tensor<int32, [5]>([-2, 1, 2, 0, 4])];
tensor<int32, [1]> var_1025_axes_0 = const()[name = tensor<string, []>("op_1025_axes_0"), val = tensor<int32, [1]>([-2])];
tensor<fp16, [3, 77, 1, 1, 512]> var_1024_cast_fp16 = transpose(perm = var_1024_perm_0, x = var_1023_cast_fp16)[name = tensor<string, []>("transpose_21")];
tensor<fp16, [3, 77, 1, 512]> var_1025_cast_fp16 = squeeze(axes = var_1025_axes_0, x = var_1024_cast_fp16)[name = tensor<string, []>("op_1025_cast_fp16")];
tensor<int32, [4]> q_55_begin_0 = const()[name = tensor<string, []>("q_55_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
tensor<int32, [4]> q_55_end_0 = const()[name = tensor<string, []>("q_55_end_0"), val = tensor<int32, [4]>([1, 77, 1, 512])];
tensor<bool, [4]> q_55_end_mask_0 = const()[name = tensor<string, []>("q_55_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> q_55_squeeze_mask_0 = const()[name = tensor<string, []>("q_55_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> q_55_cast_fp16 = slice_by_index(begin = q_55_begin_0, end = q_55_end_0, end_mask = q_55_end_mask_0, squeeze_mask = q_55_squeeze_mask_0, x = var_1025_cast_fp16)[name = tensor<string, []>("q_55_cast_fp16")];
tensor<int32, [4]> k_55_begin_0 = const()[name = tensor<string, []>("k_55_begin_0"), val = tensor<int32, [4]>([1, 0, 0, 0])];
tensor<int32, [4]> k_55_end_0 = const()[name = tensor<string, []>("k_55_end_0"), val = tensor<int32, [4]>([2, 77, 1, 512])];
tensor<bool, [4]> k_55_end_mask_0 = const()[name = tensor<string, []>("k_55_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> k_55_squeeze_mask_0 = const()[name = tensor<string, []>("k_55_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> k_55_cast_fp16 = slice_by_index(begin = k_55_begin_0, end = k_55_end_0, end_mask = k_55_end_mask_0, squeeze_mask = k_55_squeeze_mask_0, x = var_1025_cast_fp16)[name = tensor<string, []>("k_55_cast_fp16")];
tensor<int32, [4]> v_55_begin_0 = const()[name = tensor<string, []>("v_55_begin_0"), val = tensor<int32, [4]>([2, 0, 0, 0])];
tensor<int32, [4]> v_55_end_0 = const()[name = tensor<string, []>("v_55_end_0"), val = tensor<int32, [4]>([3, 77, 1, 512])];
tensor<bool, [4]> v_55_end_mask_0 = const()[name = tensor<string, []>("v_55_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> v_55_squeeze_mask_0 = const()[name = tensor<string, []>("v_55_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> v_55_cast_fp16 = slice_by_index(begin = v_55_begin_0, end = v_55_end_0, end_mask = v_55_end_mask_0, squeeze_mask = v_55_squeeze_mask_0, x = var_1025_cast_fp16)[name = tensor<string, []>("v_55_cast_fp16")];
tensor<int32, [3]> var_1034 = const()[name = tensor<string, []>("op_1034"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_1035_cast_fp16 = reshape(shape = var_1034, x = q_55_cast_fp16)[name = tensor<string, []>("op_1035_cast_fp16")];
tensor<int32, [3]> q_57_perm_0 = const()[name = tensor<string, []>("q_57_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_1041 = const()[name = tensor<string, []>("op_1041"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_1042_cast_fp16 = reshape(shape = var_1041, x = k_55_cast_fp16)[name = tensor<string, []>("op_1042_cast_fp16")];
tensor<int32, [3]> k_57_perm_0 = const()[name = tensor<string, []>("k_57_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_1048 = const()[name = tensor<string, []>("op_1048"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_1049_cast_fp16 = reshape(shape = var_1048, x = v_55_cast_fp16)[name = tensor<string, []>("op_1049_cast_fp16")];
tensor<int32, [3]> v_57_perm_0 = const()[name = tensor<string, []>("v_57_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [4]> var_1053 = const()[name = tensor<string, []>("op_1053"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> q_57_cast_fp16 = transpose(perm = q_57_perm_0, x = var_1035_cast_fp16)[name = tensor<string, []>("transpose_20")];
tensor<fp16, [1, 8, 77, 64]> q_59_cast_fp16 = reshape(shape = var_1053, x = q_57_cast_fp16)[name = tensor<string, []>("q_59_cast_fp16")];
tensor<int32, [4]> var_1055 = const()[name = tensor<string, []>("op_1055"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> k_57_cast_fp16 = transpose(perm = k_57_perm_0, x = var_1042_cast_fp16)[name = tensor<string, []>("transpose_19")];
tensor<fp16, [1, 8, 77, 64]> k_59_cast_fp16 = reshape(shape = var_1055, x = k_57_cast_fp16)[name = tensor<string, []>("k_59_cast_fp16")];
tensor<int32, [4]> var_1057 = const()[name = tensor<string, []>("op_1057"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> v_57_cast_fp16 = transpose(perm = v_57_perm_0, x = var_1049_cast_fp16)[name = tensor<string, []>("transpose_18")];
tensor<fp16, [1, 8, 77, 64]> v_59_cast_fp16 = reshape(shape = var_1057, x = v_57_cast_fp16)[name = tensor<string, []>("v_59_cast_fp16")];
tensor<fp16, []> mul_19_y_0_to_fp16 = const()[name = tensor<string, []>("mul_19_y_0_to_fp16"), val = tensor<fp16, []>(0x1p-3)];
tensor<fp16, [1, 8, 77, 64]> mul_19_cast_fp16 = mul(x = q_59_cast_fp16, y = mul_19_y_0_to_fp16)[name = tensor<string, []>("mul_19_cast_fp16")];
tensor<bool, []> matmul_9_transpose_y_0 = const()[name = tensor<string, []>("matmul_9_transpose_y_0"), val = tensor<bool, []>(true)];
tensor<bool, []> matmul_9_transpose_x_0 = const()[name = tensor<string, []>("matmul_9_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 77]> matmul_9_cast_fp16 = matmul(transpose_x = matmul_9_transpose_x_0, transpose_y = matmul_9_transpose_y_0, x = mul_19_cast_fp16, y = k_59_cast_fp16)[name = tensor<string, []>("matmul_9_cast_fp16")];
tensor<fp16, [1, 8, 77, 77]> add_9_cast_fp16 = add(x = matmul_9_cast_fp16, y = attn_mask_7_to_fp16)[name = tensor<string, []>("add_9_cast_fp16")];
tensor<int32, []> softmax_9_axis_0 = const()[name = tensor<string, []>("softmax_9_axis_0"), val = tensor<int32, []>(-1)];
tensor<fp16, [1, 8, 77, 77]> softmax_9_cast_fp16 = softmax(axis = softmax_9_axis_0, x = add_9_cast_fp16)[name = tensor<string, []>("softmax_9_cast_fp16")];
tensor<bool, []> attn_output_73_transpose_x_0 = const()[name = tensor<string, []>("attn_output_73_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<bool, []> attn_output_73_transpose_y_0 = const()[name = tensor<string, []>("attn_output_73_transpose_y_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 64]> attn_output_73_cast_fp16 = matmul(transpose_x = attn_output_73_transpose_x_0, transpose_y = attn_output_73_transpose_y_0, x = softmax_9_cast_fp16, y = v_59_cast_fp16)[name = tensor<string, []>("attn_output_73_cast_fp16")];
tensor<int32, [4]> var_1060 = const()[name = tensor<string, []>("op_1060"), val = tensor<int32, [4]>([2, 0, 1, 3])];
tensor<int32, [2]> var_1065 = const()[name = tensor<string, []>("op_1065"), val = tensor<int32, [2]>([77, 512])];
tensor<fp16, [77, 1, 8, 64]> var_1061_cast_fp16 = transpose(perm = var_1060, x = attn_output_73_cast_fp16)[name = tensor<string, []>("transpose_17")];
tensor<fp16, [77, 512]> attn_output_75_cast_fp16 = reshape(shape = var_1065, x = var_1061_cast_fp16)[name = tensor<string, []>("attn_output_75_cast_fp16")];
tensor<fp16, [512, 512]> model_transformer_resblocks_9_attn_out_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_9_attn_out_proj_weight_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(109012864)))];
tensor<fp16, [512]> model_transformer_resblocks_9_attn_out_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_9_attn_out_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(109537216)))];
tensor<fp16, [77, 512]> linear_37_cast_fp16 = linear(bias = model_transformer_resblocks_9_attn_out_proj_bias_to_fp16, weight = model_transformer_resblocks_9_attn_out_proj_weight_to_fp16, x = attn_output_75_cast_fp16)[name = tensor<string, []>("linear_37_cast_fp16")];
tensor<int32, [3]> var_1069 = const()[name = tensor<string, []>("op_1069"), val = tensor<int32, [3]>([77, 1, 512])];
tensor<fp16, [77, 1, 512]> attn_output_79_cast_fp16 = reshape(shape = var_1069, x = linear_37_cast_fp16)[name = tensor<string, []>("attn_output_79_cast_fp16")];
tensor<int32, [3]> var_1071_perm_0 = const()[name = tensor<string, []>("op_1071_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1, 77, 512]> var_1071_cast_fp16 = transpose(perm = var_1071_perm_0, x = attn_output_79_cast_fp16)[name = tensor<string, []>("transpose_16")];
tensor<fp16, [1, 77, 512]> input_93_cast_fp16 = add(x = input_91_cast_fp16, y = var_1071_cast_fp16)[name = tensor<string, []>("input_93_cast_fp16")];
tensor<int32, [1]> x_41_axes_0 = const()[name = tensor<string, []>("x_41_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_9_ln_2_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_9_ln_2_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(109538304)))];
tensor<fp16, [512]> model_transformer_resblocks_9_ln_2_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_9_ln_2_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(109539392)))];
tensor<fp16, [1, 77, 512]> x_41_cast_fp16 = layer_norm(axes = x_41_axes_0, beta = model_transformer_resblocks_9_ln_2_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_9_ln_2_weight_to_fp16, x = input_93_cast_fp16)[name = tensor<string, []>("x_41_cast_fp16")];
tensor<fp16, [2048, 512]> model_transformer_resblocks_9_mlp_c_fc_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_9_mlp_c_fc_weight_to_fp16"), val = tensor<fp16, [2048, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(109540480)))];
tensor<fp16, [2048]> model_transformer_resblocks_9_mlp_c_fc_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_9_mlp_c_fc_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(111637696)))];
tensor<fp16, [1, 77, 2048]> linear_38_cast_fp16 = linear(bias = model_transformer_resblocks_9_mlp_c_fc_bias_to_fp16, weight = model_transformer_resblocks_9_mlp_c_fc_weight_to_fp16, x = x_41_cast_fp16)[name = tensor<string, []>("linear_38_cast_fp16")];
tensor<string, []> input_99_mode_0 = const()[name = tensor<string, []>("input_99_mode_0"), val = tensor<string, []>("EXACT")];
tensor<fp16, [1, 77, 2048]> input_99_cast_fp16 = gelu(mode = input_99_mode_0, x = linear_38_cast_fp16)[name = tensor<string, []>("input_99_cast_fp16")];
tensor<fp16, [512, 2048]> model_transformer_resblocks_9_mlp_c_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_9_mlp_c_proj_weight_to_fp16"), val = tensor<fp16, [512, 2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(111641856)))];
tensor<fp16, [512]> model_transformer_resblocks_9_mlp_c_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_9_mlp_c_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(113739072)))];
tensor<fp16, [1, 77, 512]> linear_39_cast_fp16 = linear(bias = model_transformer_resblocks_9_mlp_c_proj_bias_to_fp16, weight = model_transformer_resblocks_9_mlp_c_proj_weight_to_fp16, x = input_99_cast_fp16)[name = tensor<string, []>("linear_39_cast_fp16")];
tensor<fp16, [1, 77, 512]> input_101_cast_fp16 = add(x = input_93_cast_fp16, y = linear_39_cast_fp16)[name = tensor<string, []>("input_101_cast_fp16")];
tensor<int32, [1]> x_43_axes_0 = const()[name = tensor<string, []>("x_43_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_10_ln_1_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_10_ln_1_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(113740160)))];
tensor<fp16, [512]> model_transformer_resblocks_10_ln_1_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_10_ln_1_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(113741248)))];
tensor<fp16, [1, 77, 512]> x_43_cast_fp16 = layer_norm(axes = x_43_axes_0, beta = model_transformer_resblocks_10_ln_1_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_10_ln_1_weight_to_fp16, x = input_101_cast_fp16)[name = tensor<string, []>("x_43_cast_fp16")];
tensor<int32, [3]> query_43_perm_0 = const()[name = tensor<string, []>("query_43_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1536, 512]> model_transformer_resblocks_10_attn_in_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_10_attn_in_proj_weight_to_fp16"), val = tensor<fp16, [1536, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(113742336)))];
tensor<fp16, [1536]> model_transformer_resblocks_10_attn_in_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_10_attn_in_proj_bias_to_fp16"), val = tensor<fp16, [1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(115315264)))];
tensor<fp16, [77, 1, 512]> query_43_cast_fp16 = transpose(perm = query_43_perm_0, x = x_43_cast_fp16)[name = tensor<string, []>("transpose_15")];
tensor<fp16, [77, 1, 1536]> linear_40_cast_fp16 = linear(bias = model_transformer_resblocks_10_attn_in_proj_bias_to_fp16, weight = model_transformer_resblocks_10_attn_in_proj_weight_to_fp16, x = query_43_cast_fp16)[name = tensor<string, []>("linear_40_cast_fp16")];
tensor<int32, [4]> concat_10 = const()[name = tensor<string, []>("concat_10"), val = tensor<int32, [4]>([77, 1, 3, 512])];
tensor<fp16, [77, 1, 3, 512]> var_1124_cast_fp16 = reshape(shape = concat_10, x = linear_40_cast_fp16)[name = tensor<string, []>("op_1124_cast_fp16")];
tensor<int32, [1]> var_1125_axes_0 = const()[name = tensor<string, []>("op_1125_axes_0"), val = tensor<int32, [1]>([0])];
tensor<fp16, [1, 77, 1, 3, 512]> var_1125_cast_fp16 = expand_dims(axes = var_1125_axes_0, x = var_1124_cast_fp16)[name = tensor<string, []>("op_1125_cast_fp16")];
tensor<int32, [5]> var_1126_perm_0 = const()[name = tensor<string, []>("op_1126_perm_0"), val = tensor<int32, [5]>([-2, 1, 2, 0, 4])];
tensor<int32, [1]> var_1127_axes_0 = const()[name = tensor<string, []>("op_1127_axes_0"), val = tensor<int32, [1]>([-2])];
tensor<fp16, [3, 77, 1, 1, 512]> var_1126_cast_fp16 = transpose(perm = var_1126_perm_0, x = var_1125_cast_fp16)[name = tensor<string, []>("transpose_14")];
tensor<fp16, [3, 77, 1, 512]> var_1127_cast_fp16 = squeeze(axes = var_1127_axes_0, x = var_1126_cast_fp16)[name = tensor<string, []>("op_1127_cast_fp16")];
tensor<int32, [4]> q_61_begin_0 = const()[name = tensor<string, []>("q_61_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
tensor<int32, [4]> q_61_end_0 = const()[name = tensor<string, []>("q_61_end_0"), val = tensor<int32, [4]>([1, 77, 1, 512])];
tensor<bool, [4]> q_61_end_mask_0 = const()[name = tensor<string, []>("q_61_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> q_61_squeeze_mask_0 = const()[name = tensor<string, []>("q_61_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> q_61_cast_fp16 = slice_by_index(begin = q_61_begin_0, end = q_61_end_0, end_mask = q_61_end_mask_0, squeeze_mask = q_61_squeeze_mask_0, x = var_1127_cast_fp16)[name = tensor<string, []>("q_61_cast_fp16")];
tensor<int32, [4]> k_61_begin_0 = const()[name = tensor<string, []>("k_61_begin_0"), val = tensor<int32, [4]>([1, 0, 0, 0])];
tensor<int32, [4]> k_61_end_0 = const()[name = tensor<string, []>("k_61_end_0"), val = tensor<int32, [4]>([2, 77, 1, 512])];
tensor<bool, [4]> k_61_end_mask_0 = const()[name = tensor<string, []>("k_61_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> k_61_squeeze_mask_0 = const()[name = tensor<string, []>("k_61_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> k_61_cast_fp16 = slice_by_index(begin = k_61_begin_0, end = k_61_end_0, end_mask = k_61_end_mask_0, squeeze_mask = k_61_squeeze_mask_0, x = var_1127_cast_fp16)[name = tensor<string, []>("k_61_cast_fp16")];
tensor<int32, [4]> v_61_begin_0 = const()[name = tensor<string, []>("v_61_begin_0"), val = tensor<int32, [4]>([2, 0, 0, 0])];
tensor<int32, [4]> v_61_end_0 = const()[name = tensor<string, []>("v_61_end_0"), val = tensor<int32, [4]>([3, 77, 1, 512])];
tensor<bool, [4]> v_61_end_mask_0 = const()[name = tensor<string, []>("v_61_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> v_61_squeeze_mask_0 = const()[name = tensor<string, []>("v_61_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> v_61_cast_fp16 = slice_by_index(begin = v_61_begin_0, end = v_61_end_0, end_mask = v_61_end_mask_0, squeeze_mask = v_61_squeeze_mask_0, x = var_1127_cast_fp16)[name = tensor<string, []>("v_61_cast_fp16")];
tensor<int32, [3]> var_1136 = const()[name = tensor<string, []>("op_1136"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_1137_cast_fp16 = reshape(shape = var_1136, x = q_61_cast_fp16)[name = tensor<string, []>("op_1137_cast_fp16")];
tensor<int32, [3]> q_63_perm_0 = const()[name = tensor<string, []>("q_63_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_1143 = const()[name = tensor<string, []>("op_1143"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_1144_cast_fp16 = reshape(shape = var_1143, x = k_61_cast_fp16)[name = tensor<string, []>("op_1144_cast_fp16")];
tensor<int32, [3]> k_63_perm_0 = const()[name = tensor<string, []>("k_63_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_1150 = const()[name = tensor<string, []>("op_1150"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_1151_cast_fp16 = reshape(shape = var_1150, x = v_61_cast_fp16)[name = tensor<string, []>("op_1151_cast_fp16")];
tensor<int32, [3]> v_63_perm_0 = const()[name = tensor<string, []>("v_63_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [4]> var_1155 = const()[name = tensor<string, []>("op_1155"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> q_63_cast_fp16 = transpose(perm = q_63_perm_0, x = var_1137_cast_fp16)[name = tensor<string, []>("transpose_13")];
tensor<fp16, [1, 8, 77, 64]> q_65_cast_fp16 = reshape(shape = var_1155, x = q_63_cast_fp16)[name = tensor<string, []>("q_65_cast_fp16")];
tensor<int32, [4]> var_1157 = const()[name = tensor<string, []>("op_1157"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> k_63_cast_fp16 = transpose(perm = k_63_perm_0, x = var_1144_cast_fp16)[name = tensor<string, []>("transpose_12")];
tensor<fp16, [1, 8, 77, 64]> k_65_cast_fp16 = reshape(shape = var_1157, x = k_63_cast_fp16)[name = tensor<string, []>("k_65_cast_fp16")];
tensor<int32, [4]> var_1159 = const()[name = tensor<string, []>("op_1159"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> v_63_cast_fp16 = transpose(perm = v_63_perm_0, x = var_1151_cast_fp16)[name = tensor<string, []>("transpose_11")];
tensor<fp16, [1, 8, 77, 64]> v_65_cast_fp16 = reshape(shape = var_1159, x = v_63_cast_fp16)[name = tensor<string, []>("v_65_cast_fp16")];
tensor<fp16, []> mul_21_y_0_to_fp16 = const()[name = tensor<string, []>("mul_21_y_0_to_fp16"), val = tensor<fp16, []>(0x1p-3)];
tensor<fp16, [1, 8, 77, 64]> mul_21_cast_fp16 = mul(x = q_65_cast_fp16, y = mul_21_y_0_to_fp16)[name = tensor<string, []>("mul_21_cast_fp16")];
tensor<bool, []> matmul_10_transpose_y_0 = const()[name = tensor<string, []>("matmul_10_transpose_y_0"), val = tensor<bool, []>(true)];
tensor<bool, []> matmul_10_transpose_x_0 = const()[name = tensor<string, []>("matmul_10_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 77]> matmul_10_cast_fp16 = matmul(transpose_x = matmul_10_transpose_x_0, transpose_y = matmul_10_transpose_y_0, x = mul_21_cast_fp16, y = k_65_cast_fp16)[name = tensor<string, []>("matmul_10_cast_fp16")];
tensor<fp16, [1, 8, 77, 77]> add_10_cast_fp16 = add(x = matmul_10_cast_fp16, y = attn_mask_7_to_fp16)[name = tensor<string, []>("add_10_cast_fp16")];
tensor<int32, []> softmax_10_axis_0 = const()[name = tensor<string, []>("softmax_10_axis_0"), val = tensor<int32, []>(-1)];
tensor<fp16, [1, 8, 77, 77]> softmax_10_cast_fp16 = softmax(axis = softmax_10_axis_0, x = add_10_cast_fp16)[name = tensor<string, []>("softmax_10_cast_fp16")];
tensor<bool, []> attn_output_81_transpose_x_0 = const()[name = tensor<string, []>("attn_output_81_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<bool, []> attn_output_81_transpose_y_0 = const()[name = tensor<string, []>("attn_output_81_transpose_y_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 64]> attn_output_81_cast_fp16 = matmul(transpose_x = attn_output_81_transpose_x_0, transpose_y = attn_output_81_transpose_y_0, x = softmax_10_cast_fp16, y = v_65_cast_fp16)[name = tensor<string, []>("attn_output_81_cast_fp16")];
tensor<int32, [4]> var_1162 = const()[name = tensor<string, []>("op_1162"), val = tensor<int32, [4]>([2, 0, 1, 3])];
tensor<int32, [2]> var_1167 = const()[name = tensor<string, []>("op_1167"), val = tensor<int32, [2]>([77, 512])];
tensor<fp16, [77, 1, 8, 64]> var_1163_cast_fp16 = transpose(perm = var_1162, x = attn_output_81_cast_fp16)[name = tensor<string, []>("transpose_10")];
tensor<fp16, [77, 512]> attn_output_83_cast_fp16 = reshape(shape = var_1167, x = var_1163_cast_fp16)[name = tensor<string, []>("attn_output_83_cast_fp16")];
tensor<fp16, [512, 512]> model_transformer_resblocks_10_attn_out_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_10_attn_out_proj_weight_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(115318400)))];
tensor<fp16, [512]> model_transformer_resblocks_10_attn_out_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_10_attn_out_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(115842752)))];
tensor<fp16, [77, 512]> linear_41_cast_fp16 = linear(bias = model_transformer_resblocks_10_attn_out_proj_bias_to_fp16, weight = model_transformer_resblocks_10_attn_out_proj_weight_to_fp16, x = attn_output_83_cast_fp16)[name = tensor<string, []>("linear_41_cast_fp16")];
tensor<int32, [3]> var_1171 = const()[name = tensor<string, []>("op_1171"), val = tensor<int32, [3]>([77, 1, 512])];
tensor<fp16, [77, 1, 512]> attn_output_87_cast_fp16 = reshape(shape = var_1171, x = linear_41_cast_fp16)[name = tensor<string, []>("attn_output_87_cast_fp16")];
tensor<int32, [3]> var_1173_perm_0 = const()[name = tensor<string, []>("op_1173_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1, 77, 512]> var_1173_cast_fp16 = transpose(perm = var_1173_perm_0, x = attn_output_87_cast_fp16)[name = tensor<string, []>("transpose_9")];
tensor<fp16, [1, 77, 512]> input_103_cast_fp16 = add(x = input_101_cast_fp16, y = var_1173_cast_fp16)[name = tensor<string, []>("input_103_cast_fp16")];
tensor<int32, [1]> x_45_axes_0 = const()[name = tensor<string, []>("x_45_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_10_ln_2_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_10_ln_2_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(115843840)))];
tensor<fp16, [512]> model_transformer_resblocks_10_ln_2_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_10_ln_2_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(115844928)))];
tensor<fp16, [1, 77, 512]> x_45_cast_fp16 = layer_norm(axes = x_45_axes_0, beta = model_transformer_resblocks_10_ln_2_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_10_ln_2_weight_to_fp16, x = input_103_cast_fp16)[name = tensor<string, []>("x_45_cast_fp16")];
tensor<fp16, [2048, 512]> model_transformer_resblocks_10_mlp_c_fc_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_10_mlp_c_fc_weight_to_fp16"), val = tensor<fp16, [2048, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(115846016)))];
tensor<fp16, [2048]> model_transformer_resblocks_10_mlp_c_fc_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_10_mlp_c_fc_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(117943232)))];
tensor<fp16, [1, 77, 2048]> linear_42_cast_fp16 = linear(bias = model_transformer_resblocks_10_mlp_c_fc_bias_to_fp16, weight = model_transformer_resblocks_10_mlp_c_fc_weight_to_fp16, x = x_45_cast_fp16)[name = tensor<string, []>("linear_42_cast_fp16")];
tensor<string, []> input_109_mode_0 = const()[name = tensor<string, []>("input_109_mode_0"), val = tensor<string, []>("EXACT")];
tensor<fp16, [1, 77, 2048]> input_109_cast_fp16 = gelu(mode = input_109_mode_0, x = linear_42_cast_fp16)[name = tensor<string, []>("input_109_cast_fp16")];
tensor<fp16, [512, 2048]> model_transformer_resblocks_10_mlp_c_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_10_mlp_c_proj_weight_to_fp16"), val = tensor<fp16, [512, 2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(117947392)))];
tensor<fp16, [512]> model_transformer_resblocks_10_mlp_c_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_10_mlp_c_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(120044608)))];
tensor<fp16, [1, 77, 512]> linear_43_cast_fp16 = linear(bias = model_transformer_resblocks_10_mlp_c_proj_bias_to_fp16, weight = model_transformer_resblocks_10_mlp_c_proj_weight_to_fp16, x = input_109_cast_fp16)[name = tensor<string, []>("linear_43_cast_fp16")];
tensor<fp16, [1, 77, 512]> input_111_cast_fp16 = add(x = input_103_cast_fp16, y = linear_43_cast_fp16)[name = tensor<string, []>("input_111_cast_fp16")];
tensor<int32, [1]> x_47_axes_0 = const()[name = tensor<string, []>("x_47_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_11_ln_1_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_11_ln_1_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(120045696)))];
tensor<fp16, [512]> model_transformer_resblocks_11_ln_1_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_11_ln_1_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(120046784)))];
tensor<fp16, [1, 77, 512]> x_47_cast_fp16 = layer_norm(axes = x_47_axes_0, beta = model_transformer_resblocks_11_ln_1_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_11_ln_1_weight_to_fp16, x = input_111_cast_fp16)[name = tensor<string, []>("x_47_cast_fp16")];
tensor<int32, [3]> query_perm_0 = const()[name = tensor<string, []>("query_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1536, 512]> model_transformer_resblocks_11_attn_in_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_11_attn_in_proj_weight_to_fp16"), val = tensor<fp16, [1536, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(120047872)))];
tensor<fp16, [1536]> model_transformer_resblocks_11_attn_in_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_11_attn_in_proj_bias_to_fp16"), val = tensor<fp16, [1536]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(121620800)))];
tensor<fp16, [77, 1, 512]> query_cast_fp16 = transpose(perm = query_perm_0, x = x_47_cast_fp16)[name = tensor<string, []>("transpose_8")];
tensor<fp16, [77, 1, 1536]> linear_44_cast_fp16 = linear(bias = model_transformer_resblocks_11_attn_in_proj_bias_to_fp16, weight = model_transformer_resblocks_11_attn_in_proj_weight_to_fp16, x = query_cast_fp16)[name = tensor<string, []>("linear_44_cast_fp16")];
tensor<int32, [4]> concat_11 = const()[name = tensor<string, []>("concat_11"), val = tensor<int32, [4]>([77, 1, 3, 512])];
tensor<fp16, [77, 1, 3, 512]> var_1226_cast_fp16 = reshape(shape = concat_11, x = linear_44_cast_fp16)[name = tensor<string, []>("op_1226_cast_fp16")];
tensor<int32, [1]> var_1227_axes_0 = const()[name = tensor<string, []>("op_1227_axes_0"), val = tensor<int32, [1]>([0])];
tensor<fp16, [1, 77, 1, 3, 512]> var_1227_cast_fp16 = expand_dims(axes = var_1227_axes_0, x = var_1226_cast_fp16)[name = tensor<string, []>("op_1227_cast_fp16")];
tensor<int32, [5]> var_1228_perm_0 = const()[name = tensor<string, []>("op_1228_perm_0"), val = tensor<int32, [5]>([-2, 1, 2, 0, 4])];
tensor<int32, [1]> var_1229_axes_0 = const()[name = tensor<string, []>("op_1229_axes_0"), val = tensor<int32, [1]>([-2])];
tensor<fp16, [3, 77, 1, 1, 512]> var_1228_cast_fp16 = transpose(perm = var_1228_perm_0, x = var_1227_cast_fp16)[name = tensor<string, []>("transpose_7")];
tensor<fp16, [3, 77, 1, 512]> var_1229_cast_fp16 = squeeze(axes = var_1229_axes_0, x = var_1228_cast_fp16)[name = tensor<string, []>("op_1229_cast_fp16")];
tensor<int32, [4]> q_67_begin_0 = const()[name = tensor<string, []>("q_67_begin_0"), val = tensor<int32, [4]>([0, 0, 0, 0])];
tensor<int32, [4]> q_67_end_0 = const()[name = tensor<string, []>("q_67_end_0"), val = tensor<int32, [4]>([1, 77, 1, 512])];
tensor<bool, [4]> q_67_end_mask_0 = const()[name = tensor<string, []>("q_67_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> q_67_squeeze_mask_0 = const()[name = tensor<string, []>("q_67_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> q_67_cast_fp16 = slice_by_index(begin = q_67_begin_0, end = q_67_end_0, end_mask = q_67_end_mask_0, squeeze_mask = q_67_squeeze_mask_0, x = var_1229_cast_fp16)[name = tensor<string, []>("q_67_cast_fp16")];
tensor<int32, [4]> k_67_begin_0 = const()[name = tensor<string, []>("k_67_begin_0"), val = tensor<int32, [4]>([1, 0, 0, 0])];
tensor<int32, [4]> k_67_end_0 = const()[name = tensor<string, []>("k_67_end_0"), val = tensor<int32, [4]>([2, 77, 1, 512])];
tensor<bool, [4]> k_67_end_mask_0 = const()[name = tensor<string, []>("k_67_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> k_67_squeeze_mask_0 = const()[name = tensor<string, []>("k_67_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> k_67_cast_fp16 = slice_by_index(begin = k_67_begin_0, end = k_67_end_0, end_mask = k_67_end_mask_0, squeeze_mask = k_67_squeeze_mask_0, x = var_1229_cast_fp16)[name = tensor<string, []>("k_67_cast_fp16")];
tensor<int32, [4]> v_67_begin_0 = const()[name = tensor<string, []>("v_67_begin_0"), val = tensor<int32, [4]>([2, 0, 0, 0])];
tensor<int32, [4]> v_67_end_0 = const()[name = tensor<string, []>("v_67_end_0"), val = tensor<int32, [4]>([3, 77, 1, 512])];
tensor<bool, [4]> v_67_end_mask_0 = const()[name = tensor<string, []>("v_67_end_mask_0"), val = tensor<bool, [4]>([false, true, true, true])];
tensor<bool, [4]> v_67_squeeze_mask_0 = const()[name = tensor<string, []>("v_67_squeeze_mask_0"), val = tensor<bool, [4]>([true, false, false, false])];
tensor<fp16, [77, 1, 512]> v_67_cast_fp16 = slice_by_index(begin = v_67_begin_0, end = v_67_end_0, end_mask = v_67_end_mask_0, squeeze_mask = v_67_squeeze_mask_0, x = var_1229_cast_fp16)[name = tensor<string, []>("v_67_cast_fp16")];
tensor<int32, [3]> var_1238 = const()[name = tensor<string, []>("op_1238"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_1239_cast_fp16 = reshape(shape = var_1238, x = q_67_cast_fp16)[name = tensor<string, []>("op_1239_cast_fp16")];
tensor<int32, [3]> q_69_perm_0 = const()[name = tensor<string, []>("q_69_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_1245 = const()[name = tensor<string, []>("op_1245"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_1246_cast_fp16 = reshape(shape = var_1245, x = k_67_cast_fp16)[name = tensor<string, []>("op_1246_cast_fp16")];
tensor<int32, [3]> k_69_perm_0 = const()[name = tensor<string, []>("k_69_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [3]> var_1252 = const()[name = tensor<string, []>("op_1252"), val = tensor<int32, [3]>([77, 8, 64])];
tensor<fp16, [77, 8, 64]> var_1253_cast_fp16 = reshape(shape = var_1252, x = v_67_cast_fp16)[name = tensor<string, []>("op_1253_cast_fp16")];
tensor<int32, [3]> v_69_perm_0 = const()[name = tensor<string, []>("v_69_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<int32, [4]> var_1257 = const()[name = tensor<string, []>("op_1257"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> q_69_cast_fp16 = transpose(perm = q_69_perm_0, x = var_1239_cast_fp16)[name = tensor<string, []>("transpose_6")];
tensor<fp16, [1, 8, 77, 64]> q_cast_fp16 = reshape(shape = var_1257, x = q_69_cast_fp16)[name = tensor<string, []>("q_cast_fp16")];
tensor<int32, [4]> var_1259 = const()[name = tensor<string, []>("op_1259"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> k_69_cast_fp16 = transpose(perm = k_69_perm_0, x = var_1246_cast_fp16)[name = tensor<string, []>("transpose_5")];
tensor<fp16, [1, 8, 77, 64]> k_cast_fp16 = reshape(shape = var_1259, x = k_69_cast_fp16)[name = tensor<string, []>("k_cast_fp16")];
tensor<int32, [4]> var_1261 = const()[name = tensor<string, []>("op_1261"), val = tensor<int32, [4]>([1, 8, 77, 64])];
tensor<fp16, [8, 77, 64]> v_69_cast_fp16 = transpose(perm = v_69_perm_0, x = var_1253_cast_fp16)[name = tensor<string, []>("transpose_4")];
tensor<fp16, [1, 8, 77, 64]> v_cast_fp16 = reshape(shape = var_1261, x = v_69_cast_fp16)[name = tensor<string, []>("v_cast_fp16")];
tensor<fp16, []> mul_23_y_0_to_fp16 = const()[name = tensor<string, []>("mul_23_y_0_to_fp16"), val = tensor<fp16, []>(0x1p-3)];
tensor<fp16, [1, 8, 77, 64]> mul_23_cast_fp16 = mul(x = q_cast_fp16, y = mul_23_y_0_to_fp16)[name = tensor<string, []>("mul_23_cast_fp16")];
tensor<bool, []> matmul_11_transpose_y_0 = const()[name = tensor<string, []>("matmul_11_transpose_y_0"), val = tensor<bool, []>(true)];
tensor<bool, []> matmul_11_transpose_x_0 = const()[name = tensor<string, []>("matmul_11_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 77]> matmul_11_cast_fp16 = matmul(transpose_x = matmul_11_transpose_x_0, transpose_y = matmul_11_transpose_y_0, x = mul_23_cast_fp16, y = k_cast_fp16)[name = tensor<string, []>("matmul_11_cast_fp16")];
tensor<fp16, [1, 8, 77, 77]> add_11_cast_fp16 = add(x = matmul_11_cast_fp16, y = attn_mask_7_to_fp16)[name = tensor<string, []>("add_11_cast_fp16")];
tensor<int32, []> softmax_11_axis_0 = const()[name = tensor<string, []>("softmax_11_axis_0"), val = tensor<int32, []>(-1)];
tensor<fp16, [1, 8, 77, 77]> softmax_11_cast_fp16 = softmax(axis = softmax_11_axis_0, x = add_11_cast_fp16)[name = tensor<string, []>("softmax_11_cast_fp16")];
tensor<bool, []> attn_output_89_transpose_x_0 = const()[name = tensor<string, []>("attn_output_89_transpose_x_0"), val = tensor<bool, []>(false)];
tensor<bool, []> attn_output_89_transpose_y_0 = const()[name = tensor<string, []>("attn_output_89_transpose_y_0"), val = tensor<bool, []>(false)];
tensor<fp16, [1, 8, 77, 64]> attn_output_89_cast_fp16 = matmul(transpose_x = attn_output_89_transpose_x_0, transpose_y = attn_output_89_transpose_y_0, x = softmax_11_cast_fp16, y = v_cast_fp16)[name = tensor<string, []>("attn_output_89_cast_fp16")];
tensor<int32, [4]> var_1264 = const()[name = tensor<string, []>("op_1264"), val = tensor<int32, [4]>([2, 0, 1, 3])];
tensor<int32, [2]> var_1269 = const()[name = tensor<string, []>("op_1269"), val = tensor<int32, [2]>([77, 512])];
tensor<fp16, [77, 1, 8, 64]> var_1265_cast_fp16 = transpose(perm = var_1264, x = attn_output_89_cast_fp16)[name = tensor<string, []>("transpose_3")];
tensor<fp16, [77, 512]> attn_output_91_cast_fp16 = reshape(shape = var_1269, x = var_1265_cast_fp16)[name = tensor<string, []>("attn_output_91_cast_fp16")];
tensor<fp16, [512, 512]> model_transformer_resblocks_11_attn_out_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_11_attn_out_proj_weight_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(121623936)))];
tensor<fp16, [512]> model_transformer_resblocks_11_attn_out_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_11_attn_out_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(122148288)))];
tensor<fp16, [77, 512]> linear_45_cast_fp16 = linear(bias = model_transformer_resblocks_11_attn_out_proj_bias_to_fp16, weight = model_transformer_resblocks_11_attn_out_proj_weight_to_fp16, x = attn_output_91_cast_fp16)[name = tensor<string, []>("linear_45_cast_fp16")];
tensor<int32, [3]> var_1273 = const()[name = tensor<string, []>("op_1273"), val = tensor<int32, [3]>([77, 1, 512])];
tensor<fp16, [77, 1, 512]> attn_output_cast_fp16 = reshape(shape = var_1273, x = linear_45_cast_fp16)[name = tensor<string, []>("attn_output_cast_fp16")];
tensor<int32, [3]> var_1275_perm_0 = const()[name = tensor<string, []>("op_1275_perm_0"), val = tensor<int32, [3]>([1, 0, 2])];
tensor<fp16, [1, 77, 512]> var_1275_cast_fp16 = transpose(perm = var_1275_perm_0, x = attn_output_cast_fp16)[name = tensor<string, []>("transpose_2")];
tensor<fp16, [1, 77, 512]> input_113_cast_fp16 = add(x = input_111_cast_fp16, y = var_1275_cast_fp16)[name = tensor<string, []>("input_113_cast_fp16")];
tensor<int32, [1]> x_49_axes_0 = const()[name = tensor<string, []>("x_49_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_transformer_resblocks_11_ln_2_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_11_ln_2_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(122149376)))];
tensor<fp16, [512]> model_transformer_resblocks_11_ln_2_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_11_ln_2_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(122150464)))];
tensor<fp16, [1, 77, 512]> x_49_cast_fp16 = layer_norm(axes = x_49_axes_0, beta = model_transformer_resblocks_11_ln_2_bias_to_fp16, epsilon = var_34_to_fp16, gamma = model_transformer_resblocks_11_ln_2_weight_to_fp16, x = input_113_cast_fp16)[name = tensor<string, []>("x_49_cast_fp16")];
tensor<fp16, [2048, 512]> model_transformer_resblocks_11_mlp_c_fc_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_11_mlp_c_fc_weight_to_fp16"), val = tensor<fp16, [2048, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(122151552)))];
tensor<fp16, [2048]> model_transformer_resblocks_11_mlp_c_fc_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_11_mlp_c_fc_bias_to_fp16"), val = tensor<fp16, [2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(124248768)))];
tensor<fp16, [1, 77, 2048]> linear_46_cast_fp16 = linear(bias = model_transformer_resblocks_11_mlp_c_fc_bias_to_fp16, weight = model_transformer_resblocks_11_mlp_c_fc_weight_to_fp16, x = x_49_cast_fp16)[name = tensor<string, []>("linear_46_cast_fp16")];
tensor<string, []> input_119_mode_0 = const()[name = tensor<string, []>("input_119_mode_0"), val = tensor<string, []>("EXACT")];
tensor<fp16, [1, 77, 2048]> input_119_cast_fp16 = gelu(mode = input_119_mode_0, x = linear_46_cast_fp16)[name = tensor<string, []>("input_119_cast_fp16")];
tensor<fp16, [512, 2048]> model_transformer_resblocks_11_mlp_c_proj_weight_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_11_mlp_c_proj_weight_to_fp16"), val = tensor<fp16, [512, 2048]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(124252928)))];
tensor<fp16, [512]> model_transformer_resblocks_11_mlp_c_proj_bias_to_fp16 = const()[name = tensor<string, []>("model_transformer_resblocks_11_mlp_c_proj_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(126350144)))];
tensor<fp16, [1, 77, 512]> linear_47_cast_fp16 = linear(bias = model_transformer_resblocks_11_mlp_c_proj_bias_to_fp16, weight = model_transformer_resblocks_11_mlp_c_proj_weight_to_fp16, x = input_119_cast_fp16)[name = tensor<string, []>("linear_47_cast_fp16")];
tensor<fp16, [1, 77, 512]> input_cast_fp16 = add(x = input_113_cast_fp16, y = linear_47_cast_fp16)[name = tensor<string, []>("input_cast_fp16")];
tensor<int32, [1]> x_51_axes_0 = const()[name = tensor<string, []>("x_51_axes_0"), val = tensor<int32, [1]>([-1])];
tensor<fp16, [512]> model_ln_final_weight_to_fp16 = const()[name = tensor<string, []>("model_ln_final_weight_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(126351232)))];
tensor<fp16, [512]> model_ln_final_bias_to_fp16 = const()[name = tensor<string, []>("model_ln_final_bias_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(126352320)))];
tensor<fp16, []> var_1296_to_fp16 = const()[name = tensor<string, []>("op_1296_to_fp16"), val = tensor<fp16, []>(0x1.5p-17)];
tensor<fp16, [1, 77, 512]> x_51_cast_fp16 = layer_norm(axes = x_51_axes_0, beta = model_ln_final_bias_to_fp16, epsilon = var_1296_to_fp16, gamma = model_ln_final_weight_to_fp16, x = input_cast_fp16)[name = tensor<string, []>("x_51_cast_fp16")];
tensor<int32, [1]> var_1311 = const()[name = tensor<string, []>("op_1311"), val = tensor<int32, [1]>([0])];
tensor<int32, []> var_1314_axis_0 = const()[name = tensor<string, []>("op_1314_axis_0"), val = tensor<int32, []>(-1)];
tensor<bool, []> var_1314_keep_dims_0 = const()[name = tensor<string, []>("op_1314_keep_dims_0"), val = tensor<bool, []>(false)];
tensor<int32, [1]> var_1314 = reduce_argmax(axis = var_1314_axis_0, keep_dims = var_1314_keep_dims_0, x = text)[name = tensor<string, []>("op_1314")];
tensor<int32, []> stack_0_axis_0 = const()[name = tensor<string, []>("stack_0_axis_0"), val = tensor<int32, []>(1)];
tensor<int32, [1, 2]> stack_0 = stack(axis = stack_0_axis_0, values = (var_1311, var_1314))[name = tensor<string, []>("stack_0")];
tensor<fp16, [1, 512]> x_transpose_cast_fp16 = gather_nd(indices = stack_0, x = x_51_cast_fp16)[name = tensor<string, []>("x_transpose_cast_fp16")];
tensor<fp16, [512, 512]> transpose_1_to_fp16 = const()[name = tensor<string, []>("transpose_1_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(126353408)))];
tensor<fp16, [512]> var_1317_bias_0_to_fp16 = const()[name = tensor<string, []>("op_1317_bias_0_to_fp16"), val = tensor<fp16, [512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(126877760)))];
tensor<fp16, [1, 512]> var_1317_cast_fp16 = linear(bias = var_1317_bias_0_to_fp16, weight = transpose_1_to_fp16, x = x_transpose_cast_fp16)[name = tensor<string, []>("op_1317_cast_fp16")];
tensor<string, []> var_1317_cast_fp16_to_fp32_dtype_0 = const()[name = tensor<string, []>("op_1317_cast_fp16_to_fp32_dtype_0"), val = tensor<string, []>("fp32")];
tensor<fp32, [1, 512]> var_1317 = cast(dtype = var_1317_cast_fp16_to_fp32_dtype_0, x = var_1317_cast_fp16)[name = tensor<string, []>("cast_133")];
} -> (var_1317);
} |