File size: 23,751 Bytes
f554427 722eedf f554427 722eedf f554427 722eedf f554427 722eedf f554427 722eedf f554427 722eedf f554427 722eedf f554427 722eedf f554427 |
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 |
program(1.3)
[buildInfo = dict<string, string>({{"coremlc-component-MIL", "3400.42.1"}, {"coremlc-version", "3400.51.1"}})]
{
func input_1_context_512<ios18>(tensor<int32, [1]> full_sequence_length, tensor<int32, [1, 4]> input_ids) {
tensor<int32, [1]> T = const()[name = string("T"), val = tensor<int32, [1]>([4])];
int32 x_axis_0 = const()[name = string("x_axis_0"), val = int32(0)];
int32 x_batch_dims_0 = const()[name = string("x_batch_dims_0"), val = int32(0)];
bool x_validate_indices_0 = const()[name = string("x_validate_indices_0"), val = bool(false)];
tensor<fp16, [32000, 4096]> wte_weight_to_fp16 = const()[name = string("wte_weight_to_fp16"), val = tensor<fp16, [32000, 4096]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(64)))];
string input_ids_to_int16_dtype_0 = const()[name = string("input_ids_to_int16_dtype_0"), val = string("int16")];
tensor<int16, [1, 4]> input_ids_to_int16 = cast(dtype = input_ids_to_int16_dtype_0, x = input_ids)[name = string("cast_6")];
tensor<fp16, [1, 4, 4096]> x_cast_fp16_cast_uint16 = gather(axis = x_axis_0, batch_dims = x_batch_dims_0, indices = input_ids_to_int16, validate_indices = x_validate_indices_0, x = wte_weight_to_fp16)[name = string("x_cast_fp16_cast_uint16")];
tensor<int32, [3]> var_16_perm_0 = const()[name = string("op_16_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
tensor<int32, [1]> var_18_axes_0 = const()[name = string("op_18_axes_0"), val = tensor<int32, [1]>([2])];
tensor<fp16, [1, 4096, 4]> var_16_cast_fp16 = transpose(perm = var_16_perm_0, x = x_cast_fp16_cast_uint16)[name = string("transpose_0")];
tensor<fp16, [1, 4096, 1, 4]> x = expand_dims(axes = var_18_axes_0, x = var_16_cast_fp16)[name = string("op_18_cast_fp16")];
tensor<int32, [1]> pos_offset = sub(x = T, y = full_sequence_length)[name = string("pos_offset")];
tensor<int32, [4]> var_26 = const()[name = string("op_26"), val = tensor<int32, [4]>([0, 1, 2, 3])];
tensor<int32, [4]> input_pos_1 = sub(x = var_26, y = pos_offset)[name = string("input_pos_1")];
tensor<int32, [4]> var_34 = const()[name = string("op_34"), val = tensor<int32, [4]>([0, 0, 0, 0])];
tensor<int32, [4]> input_pos = maximum(x = input_pos_1, y = var_34)[name = string("input_pos")];
int32 var_45 = const()[name = string("op_45"), val = int32(1)];
int32 var_46_batch_dims_0 = const()[name = string("op_46_batch_dims_0"), val = int32(0)];
bool var_46_validate_indices_0 = const()[name = string("op_46_validate_indices_0"), val = bool(false)];
tensor<fp16, [128, 512]> var_44_to_fp16 = const()[name = string("op_44_to_fp16"), val = tensor<fp16, [128, 512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(262144128)))];
string input_pos_to_uint16_dtype_0 = const()[name = string("input_pos_to_uint16_dtype_0"), val = string("uint16")];
tensor<uint16, [4]> input_pos_to_uint16 = cast(dtype = input_pos_to_uint16_dtype_0, x = input_pos)[name = string("cast_5")];
tensor<fp16, [128, 4]> cos = gather(axis = var_45, batch_dims = var_46_batch_dims_0, indices = input_pos_to_uint16, validate_indices = var_46_validate_indices_0, x = var_44_to_fp16)[name = string("op_46_cast_fp16_cast_uint16")];
int32 var_56 = const()[name = string("op_56"), val = int32(1)];
int32 var_57_batch_dims_0 = const()[name = string("op_57_batch_dims_0"), val = int32(0)];
bool var_57_validate_indices_0 = const()[name = string("op_57_validate_indices_0"), val = bool(false)];
tensor<fp16, [128, 512]> var_55_to_fp16 = const()[name = string("op_55_to_fp16"), val = tensor<fp16, [128, 512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(262275264)))];
tensor<fp16, [128, 4]> sin = gather(axis = var_56, batch_dims = var_57_batch_dims_0, indices = input_pos_to_uint16, validate_indices = var_57_validate_indices_0, x = var_55_to_fp16)[name = string("op_57_cast_fp16_cast_uint16")];
tensor<int32, [4, 1]> var_92 = const()[name = string("op_92"), val = tensor<int32, [4, 1]>([[0], [1], [2], [3]])];
tensor<bool, [4, 1]> var_95 = less(x = var_92, y = pos_offset)[name = string("op_95")];
tensor<int32, [2]> var_95_after_broadcast_reps_0 = const()[name = string("op_95_after_broadcast_reps_0"), val = tensor<int32, [2]>([1, 512])];
tensor<bool, [4, 512]> var_95_after_broadcast = tile(reps = var_95_after_broadcast_reps_0, x = var_95)[name = string("op_95_after_broadcast")];
tensor<fp16, [4, 512]> all_mask_to_fp16 = const()[name = string("all_mask_to_fp16"), val = tensor<fp16, [4, 512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(263455104)))];
tensor<fp16, [4, 512]> m_1_to_fp16 = const()[name = string("m_1_to_fp16"), val = tensor<fp16, [4, 512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(263459264)))];
tensor<fp16, [4, 512]> m_3_cast_fp16 = select(a = all_mask_to_fp16, b = m_1_to_fp16, cond = var_95_after_broadcast)[name = string("m_3_cast_fp16")];
tensor<int32, [512]> var_105 = const()[name = string("op_105"), val = tensor<int32, [512]>([0, 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])];
int32 var_106 = const()[name = string("op_106"), val = int32(512)];
tensor<int32, [1]> var_108 = sub(x = var_106, y = full_sequence_length)[name = string("op_108")];
tensor<bool, [512]> var_109 = less(x = var_105, y = var_108)[name = string("op_109")];
tensor<int32, [1]> expand_dims_0_axes_0 = const()[name = string("expand_dims_0_axes_0"), val = tensor<int32, [1]>([0])];
tensor<bool, [1, 512]> expand_dims_0 = expand_dims(axes = expand_dims_0_axes_0, x = var_109)[name = string("expand_dims_0")];
tensor<int32, [2]> var_109_after_broadcast_reps_0 = const()[name = string("op_109_after_broadcast_reps_0"), val = tensor<int32, [2]>([4, 1])];
tensor<bool, [4, 512]> var_109_after_broadcast = tile(reps = var_109_after_broadcast_reps_0, x = expand_dims_0)[name = string("op_109_after_broadcast")];
tensor<fp16, [4, 512]> m_cast_fp16 = select(a = all_mask_to_fp16, b = m_3_cast_fp16, cond = var_109_after_broadcast)[name = string("m_cast_fp16")];
tensor<int32, [1]> var_112_axes_0 = const()[name = string("op_112_axes_0"), val = tensor<int32, [1]>([0])];
tensor<fp16, [1, 4, 512]> var_112_cast_fp16 = expand_dims(axes = var_112_axes_0, x = m_cast_fp16)[name = string("op_112_cast_fp16")];
tensor<int32, [1]> var_114_axes_0 = const()[name = string("op_114_axes_0"), val = tensor<int32, [1]>([0])];
tensor<fp16, [1, 1, 4, 512]> mask = expand_dims(axes = var_114_axes_0, x = var_112_cast_fp16)[name = string("op_114_cast_fp16")];
} -> (x, cos, sin, mask);
func input_512_context_512<ios18>(tensor<int32, [1]> full_sequence_length, tensor<int32, [1, 512]> input_ids) {
tensor<int32, [1]> T = const()[name = string("T"), val = tensor<int32, [1]>([512])];
int32 x_axis_0 = const()[name = string("x_axis_0"), val = int32(0)];
int32 x_batch_dims_0 = const()[name = string("x_batch_dims_0"), val = int32(0)];
bool x_validate_indices_0 = const()[name = string("x_validate_indices_0"), val = bool(false)];
tensor<fp16, [32000, 4096]> wte_weight_to_fp16 = const()[name = string("wte_weight_to_fp16"), val = tensor<fp16, [32000, 4096]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(64)))];
string input_ids_to_int16_dtype_0 = const()[name = string("input_ids_to_int16_dtype_0"), val = string("int16")];
tensor<int16, [1, 512]> input_ids_to_int16 = cast(dtype = input_ids_to_int16_dtype_0, x = input_ids)[name = string("cast_6")];
tensor<fp16, [1, 512, 4096]> x_cast_fp16_cast_uint16 = gather(axis = x_axis_0, batch_dims = x_batch_dims_0, indices = input_ids_to_int16, validate_indices = x_validate_indices_0, x = wte_weight_to_fp16)[name = string("x_cast_fp16_cast_uint16")];
tensor<int32, [3]> var_16_perm_0 = const()[name = string("op_16_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
tensor<int32, [1]> var_18_axes_0 = const()[name = string("op_18_axes_0"), val = tensor<int32, [1]>([2])];
tensor<fp16, [1, 4096, 512]> var_16_cast_fp16 = transpose(perm = var_16_perm_0, x = x_cast_fp16_cast_uint16)[name = string("transpose_0")];
tensor<fp16, [1, 4096, 1, 512]> x = expand_dims(axes = var_18_axes_0, x = var_16_cast_fp16)[name = string("op_18_cast_fp16")];
tensor<int32, [1]> pos_offset = sub(x = T, y = full_sequence_length)[name = string("pos_offset")];
tensor<int32, [512]> var_26 = const()[name = string("op_26"), val = tensor<int32, [512]>([0, 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])];
tensor<int32, [512]> input_pos_1 = sub(x = var_26, y = pos_offset)[name = string("input_pos_1")];
tensor<int32, [512]> var_34 = const()[name = string("op_34"), val = tensor<int32, [512]>([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0])];
tensor<int32, [512]> input_pos = maximum(x = input_pos_1, y = var_34)[name = string("input_pos")];
int32 var_45 = const()[name = string("op_45"), val = int32(1)];
int32 var_46_batch_dims_0 = const()[name = string("op_46_batch_dims_0"), val = int32(0)];
bool var_46_validate_indices_0 = const()[name = string("op_46_validate_indices_0"), val = bool(false)];
tensor<fp16, [128, 512]> var_44_to_fp16 = const()[name = string("op_44_to_fp16"), val = tensor<fp16, [128, 512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(262144128)))];
string input_pos_to_uint16_dtype_0 = const()[name = string("input_pos_to_uint16_dtype_0"), val = string("uint16")];
tensor<uint16, [512]> input_pos_to_uint16 = cast(dtype = input_pos_to_uint16_dtype_0, x = input_pos)[name = string("cast_5")];
tensor<fp16, [128, 512]> cos = gather(axis = var_45, batch_dims = var_46_batch_dims_0, indices = input_pos_to_uint16, validate_indices = var_46_validate_indices_0, x = var_44_to_fp16)[name = string("op_46_cast_fp16_cast_uint16")];
int32 var_56 = const()[name = string("op_56"), val = int32(1)];
int32 var_57_batch_dims_0 = const()[name = string("op_57_batch_dims_0"), val = int32(0)];
bool var_57_validate_indices_0 = const()[name = string("op_57_validate_indices_0"), val = bool(false)];
tensor<fp16, [128, 512]> var_55_to_fp16 = const()[name = string("op_55_to_fp16"), val = tensor<fp16, [128, 512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(262275264)))];
tensor<fp16, [128, 512]> sin = gather(axis = var_56, batch_dims = var_57_batch_dims_0, indices = input_pos_to_uint16, validate_indices = var_57_validate_indices_0, x = var_55_to_fp16)[name = string("op_57_cast_fp16_cast_uint16")];
tensor<int32, [512, 1]> var_92 = const()[name = string("op_92"), val = tensor<int32, [512, 1]>([[0], [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]])];
tensor<bool, [512, 1]> var_95 = less(x = var_92, y = pos_offset)[name = string("op_95")];
tensor<int32, [2]> var_95_after_broadcast_reps_0 = const()[name = string("op_95_after_broadcast_reps_0"), val = tensor<int32, [2]>([1, 512])];
tensor<bool, [512, 512]> var_95_after_broadcast = tile(reps = var_95_after_broadcast_reps_0, x = var_95)[name = string("op_95_after_broadcast")];
tensor<fp16, [512, 512]> all_mask_to_fp16 = const()[name = string("all_mask_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(262406400)))];
tensor<fp16, [512, 512]> m_1_to_fp16 = const()[name = string("m_1_to_fp16"), val = tensor<fp16, [512, 512]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(262930752)))];
tensor<fp16, [512, 512]> m_3_cast_fp16 = select(a = all_mask_to_fp16, b = m_1_to_fp16, cond = var_95_after_broadcast)[name = string("m_3_cast_fp16")];
int32 var_106 = const()[name = string("op_106"), val = int32(512)];
tensor<int32, [1]> var_108 = sub(x = var_106, y = full_sequence_length)[name = string("op_108")];
tensor<bool, [512]> var_109 = less(x = var_26, y = var_108)[name = string("op_109")];
tensor<int32, [1]> expand_dims_0_axes_0 = const()[name = string("expand_dims_0_axes_0"), val = tensor<int32, [1]>([0])];
tensor<bool, [1, 512]> expand_dims_0 = expand_dims(axes = expand_dims_0_axes_0, x = var_109)[name = string("expand_dims_0")];
tensor<int32, [2]> var_109_after_broadcast_reps_0 = const()[name = string("op_109_after_broadcast_reps_0"), val = tensor<int32, [2]>([512, 1])];
tensor<bool, [512, 512]> var_109_after_broadcast = tile(reps = var_109_after_broadcast_reps_0, x = expand_dims_0)[name = string("op_109_after_broadcast")];
tensor<fp16, [512, 512]> m_cast_fp16 = select(a = all_mask_to_fp16, b = m_3_cast_fp16, cond = var_109_after_broadcast)[name = string("m_cast_fp16")];
tensor<int32, [1]> var_112_axes_0 = const()[name = string("op_112_axes_0"), val = tensor<int32, [1]>([0])];
tensor<fp16, [1, 512, 512]> var_112_cast_fp16 = expand_dims(axes = var_112_axes_0, x = m_cast_fp16)[name = string("op_112_cast_fp16")];
tensor<int32, [1]> var_114_axes_0 = const()[name = string("op_114_axes_0"), val = tensor<int32, [1]>([0])];
tensor<fp16, [1, 1, 512, 512]> mask = expand_dims(axes = var_114_axes_0, x = var_112_cast_fp16)[name = string("op_114_cast_fp16")];
} -> (x, cos, sin, mask);
} |