smpanaro commited on Aug 15

Commit

dba673f

•

1 Parent(s): 722eedf

Update Sonoma model with faster 8x8 conv and split einsum attention

Browse files

This view is limited to 50 files because it contains too many changes. See raw diff

Files changed (50) hide show

Llama-2-7b-hf_chunk1.mlmodelc/analytics/coremldata.bin +1 -1
Llama-2-7b-hf_chunk1.mlmodelc/coremldata.bin +2 -2
Llama-2-7b-hf_chunk1.mlmodelc/metadata.json +9 -8
Llama-2-7b-hf_chunk1.mlmodelc/model.mil +39 -37
Llama-2-7b-hf_chunk1.mlmodelc/weights/weight.bin +1 -1
Llama-2-7b-hf_chunk10.mlmodelc/analytics/coremldata.bin +1 -1
Llama-2-7b-hf_chunk10.mlmodelc/coremldata.bin +2 -2
Llama-2-7b-hf_chunk10.mlmodelc/metadata.json +42 -41
Llama-2-7b-hf_chunk10.mlmodelc/model.mil +0 -0
Llama-2-7b-hf_chunk10.mlmodelc/weights/weight.bin +2 -2
Llama-2-7b-hf_chunk11.mlmodelc/analytics/coremldata.bin +1 -1
Llama-2-7b-hf_chunk11.mlmodelc/coremldata.bin +2 -2
Llama-2-7b-hf_chunk11.mlmodelc/metadata.json +42 -41
Llama-2-7b-hf_chunk11.mlmodelc/model.mil +0 -0
Llama-2-7b-hf_chunk11.mlmodelc/weights/weight.bin +2 -2
Llama-2-7b-hf_chunk12.mlmodelc/analytics/coremldata.bin +1 -1
Llama-2-7b-hf_chunk12.mlmodelc/coremldata.bin +2 -2
Llama-2-7b-hf_chunk12.mlmodelc/metadata.json +35 -34
Llama-2-7b-hf_chunk12.mlmodelc/model.mil +0 -0
Llama-2-7b-hf_chunk12.mlmodelc/weights/weight.bin +2 -2
Llama-2-7b-hf_chunk13.mlmodelc/analytics/coremldata.bin +1 -1
Llama-2-7b-hf_chunk13.mlmodelc/coremldata.bin +2 -2
Llama-2-7b-hf_chunk13.mlmodelc/metadata.json +10 -11
Llama-2-7b-hf_chunk13.mlmodelc/model.mil +25 -22
Llama-2-7b-hf_chunk13.mlmodelc/weights/weight.bin +2 -2
Llama-2-7b-hf_chunk2.mlmodelc/analytics/coremldata.bin +1 -1
Llama-2-7b-hf_chunk2.mlmodelc/coremldata.bin +2 -2
Llama-2-7b-hf_chunk2.mlmodelc/metadata.json +42 -41
Llama-2-7b-hf_chunk2.mlmodelc/model.mil +0 -0
Llama-2-7b-hf_chunk2.mlmodelc/weights/weight.bin +2 -2
Llama-2-7b-hf_chunk3.mlmodelc/analytics/coremldata.bin +1 -1
Llama-2-7b-hf_chunk3.mlmodelc/coremldata.bin +2 -2
Llama-2-7b-hf_chunk3.mlmodelc/metadata.json +42 -41
Llama-2-7b-hf_chunk3.mlmodelc/model.mil +0 -0
Llama-2-7b-hf_chunk3.mlmodelc/weights/weight.bin +2 -2
Llama-2-7b-hf_chunk4.mlmodelc/analytics/coremldata.bin +1 -1
Llama-2-7b-hf_chunk4.mlmodelc/coremldata.bin +2 -2
Llama-2-7b-hf_chunk4.mlmodelc/metadata.json +42 -41
Llama-2-7b-hf_chunk4.mlmodelc/model.mil +0 -0
Llama-2-7b-hf_chunk4.mlmodelc/weights/weight.bin +2 -2
Llama-2-7b-hf_chunk5.mlmodelc/analytics/coremldata.bin +1 -1
Llama-2-7b-hf_chunk5.mlmodelc/coremldata.bin +2 -2
Llama-2-7b-hf_chunk5.mlmodelc/metadata.json +42 -41
Llama-2-7b-hf_chunk5.mlmodelc/model.mil +0 -0
Llama-2-7b-hf_chunk5.mlmodelc/weights/weight.bin +2 -2
Llama-2-7b-hf_chunk6.mlmodelc/analytics/coremldata.bin +1 -1
Llama-2-7b-hf_chunk6.mlmodelc/coremldata.bin +2 -2
Llama-2-7b-hf_chunk6.mlmodelc/metadata.json +42 -41
Llama-2-7b-hf_chunk6.mlmodelc/model.mil +0 -0
Llama-2-7b-hf_chunk6.mlmodelc/weights/weight.bin +2 -2

Llama-2-7b-hf_chunk1.mlmodelc/analytics/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2184893e48a9da76b01012a32cca3e2ebfd4080553daa78318fe2391679dd7fe
 size 243

 version https://git-lfs.github.com/spec/v1
+oid sha256:75aec049f180c798ac703b21d39421b0cb60122e24f20933de1e3729242356ac
 size 243

Llama-2-7b-hf_chunk1.mlmodelc/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8dbc016e9274c2a01d9eddb55dfd163a8ae74e7e97f0932268602c1a8b14903c
-size 407

 version https://git-lfs.github.com/spec/v1
+oid sha256:8eea6be9d5db2d6fafee48085cec7ae6575566482e1f313ab8eaf35b75c0fbdf
+size 409

Llama-2-7b-hf_chunk1.mlmodelc/metadata.json CHANGED Viewed

@@ -7,9 +7,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 4096, 1, 64]",
         "name" : "x",
         "type" : "MultiArray"
       },
@@ -37,9 +37,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 1 × 64 × 512)",
         "shortDescription" : "",
-        "shape" : "[1, 1, 64, 512]",
         "name" : "mask",
         "type" : "MultiArray"
       }
@@ -52,9 +52,10 @@
       "Select" : 2,
       "Tile" : 2,
       "Ios16.sub" : 3,
-      "Transpose" : 1,
       "Ios16.gather" : 3,
-      "ExpandDims" : 4,
       "Ios16.maximum" : 1,
       "Ios16.less" : 2
     },
@@ -74,7 +75,7 @@
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
       "com.github.apple.coremltools.source" : "torch==2.1.0",
-      "com.github.apple.coremltools.version" : "7.2"
     },
     "inputSchema" : [
       {
@@ -98,7 +99,7 @@
         "type" : "MultiArray"
       }
     ],
-    "generatedClassName" : "Llama_2_7b_hf_2024_05_25_14_03_55_chunk1",
     "method" : "predict"
   }
 ]

         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 8 × 8)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 8, 8]",
         "name" : "x",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 512 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 512, 1, 64]",
         "name" : "mask",
         "type" : "MultiArray"
       }
       "Select" : 2,
       "Tile" : 2,
       "Ios16.sub" : 3,
+      "Transpose" : 2,
       "Ios16.gather" : 3,
+      "ExpandDims" : 3,
+      "Ios16.reshape" : 1,
       "Ios16.maximum" : 1,
       "Ios16.less" : 2
     },
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
       "com.github.apple.coremltools.source" : "torch==2.1.0",
+      "com.github.apple.coremltools.version" : "8.0b1"
     },
     "inputSchema" : [
       {
         "type" : "MultiArray"
       }
     ],
+    "generatedClassName" : "Llama_2_7b_hf_2024_08_09_09_54_41_chunk1",
     "method" : "predict"
   }
 ]

Llama-2-7b-hf_chunk1.mlmodelc/model.mil CHANGED Viewed

@@ -1,48 +1,50 @@
 program(1.0)
-[buildInfo = dict<tensor<string, []>, tensor<string, []>>({{"coremlc-component-MIL", "5.33.5"}, {"coremlc-version", "1877.40.3"}, {"coremltools-component-torch", "2.1.0"}, {"coremltools-source-dialect", "TorchScript"}, {"coremltools-version", "7.2"}})]
 {
     func main<ios16>(tensor<int32, [1]> full_sequence_length, tensor<int32, [1, 64]> input_ids) {
             tensor<int32, [1]> T = const()[name = tensor<string, []>("T"), val = tensor<int32, [1]>([64])];
-            tensor<int32, []> x_axis_0 = const()[name = tensor<string, []>("x_axis_0"), val = tensor<int32, []>(0)];
-            tensor<int32, []> x_batch_dims_0 = const()[name = tensor<string, []>("x_batch_dims_0"), val = tensor<int32, []>(0)];
             tensor<fp16, [32000, 4096]> wte_weight_to_fp16 = const()[name = tensor<string, []>("wte_weight_to_fp16"), val = tensor<fp16, [32000, 4096]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(64)))];
-            tensor<fp16, [1, 64, 4096]> x_cast_fp16 = gather(axis = x_axis_0, batch_dims = x_batch_dims_0, indices = input_ids, x = wte_weight_to_fp16)[name = tensor<string, []>("x_cast_fp16")];
-            tensor<int32, [3]> var_16_perm_0 = const()[name = tensor<string, []>("op_16_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
-            tensor<int32, [1]> var_18_axes_0 = const()[name = tensor<string, []>("op_18_axes_0"), val = tensor<int32, [1]>([2])];
-            tensor<fp16, [1, 4096, 64]> transpose_0 = transpose(perm = var_16_perm_0, x = x_cast_fp16)[name = tensor<string, []>("transpose_0")];
-            tensor<fp16, [1, 4096, 1, 64]> x = expand_dims(axes = var_18_axes_0, x = transpose_0)[name = tensor<string, []>("op_18_cast_fp16")];
             tensor<int32, [1]> pos_offset = sub(x = T, y = full_sequence_length)[name = tensor<string, []>("pos_offset")];
-            tensor<int32, [64]> var_26 = const()[name = tensor<string, []>("op_26"), val = tensor<int32, [64]>([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])];
-            tensor<int32, [64]> input_pos_1 = sub(x = var_26, y = pos_offset)[name = tensor<string, []>("input_pos_1")];
-            tensor<int32, [64]> var_34 = const()[name = tensor<string, []>("op_34"), val = tensor<int32, [64]>([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 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, [64]> input_pos = maximum(x = input_pos_1, y = var_34)[name = tensor<string, []>("input_pos")];
-            tensor<int32, []> var_45 = const()[name = tensor<string, []>("op_45"), val = tensor<int32, []>(1)];
-            tensor<int32, []> var_46_batch_dims_0 = const()[name = tensor<string, []>("op_46_batch_dims_0"), val = tensor<int32, []>(0)];
-            tensor<fp16, [128, 512]> var_44_to_fp16 = const()[name = tensor<string, []>("op_44_to_fp16"), val = tensor<fp16, [128, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(262144128)))];
-            tensor<fp16, [128, 64]> cos = gather(axis = var_45, batch_dims = var_46_batch_dims_0, indices = input_pos, x = var_44_to_fp16)[name = tensor<string, []>("op_46_cast_fp16")];
-            tensor<int32, []> var_56 = const()[name = tensor<string, []>("op_56"), val = tensor<int32, []>(1)];
-            tensor<int32, []> var_57_batch_dims_0 = const()[name = tensor<string, []>("op_57_batch_dims_0"), val = tensor<int32, []>(0)];
-            tensor<fp16, [128, 512]> var_55_to_fp16 = const()[name = tensor<string, []>("op_55_to_fp16"), val = tensor<fp16, [128, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(262275264)))];
-            tensor<fp16, [128, 64]> sin = gather(axis = var_56, batch_dims = var_57_batch_dims_0, indices = input_pos, x = var_55_to_fp16)[name = tensor<string, []>("op_57_cast_fp16")];
-            tensor<int32, [64, 1]> var_92 = const()[name = tensor<string, []>("op_92"), val = tensor<int32, [64, 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]])];
-            tensor<bool, [64, 1]> var_95 = less(x = var_92, y = pos_offset)[name = tensor<string, []>("op_95")];
-            tensor<int32, [2]> var_95_after_broadcast_reps_0 = const()[name = tensor<string, []>("op_95_after_broadcast_reps_0"), val = tensor<int32, [2]>([1, 512])];
-            tensor<bool, [64, 512]> var_95_after_broadcast = tile(reps = var_95_after_broadcast_reps_0, x = var_95)[name = tensor<string, []>("op_95_after_broadcast")];
             tensor<fp16, [64, 512]> all_mask_to_fp16 = const()[name = tensor<string, []>("all_mask_to_fp16"), val = tensor<fp16, [64, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(262406400)))];
             tensor<fp16, [64, 512]> m_1_to_fp16 = const()[name = tensor<string, []>("m_1_to_fp16"), val = tensor<fp16, [64, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(262472000)))];
-            tensor<fp16, [64, 512]> m_3_cast_fp16 = select(a = all_mask_to_fp16, b = m_1_to_fp16, cond = var_95_after_broadcast)[name = tensor<string, []>("m_3_cast_fp16")];
-            tensor<int32, [512]> var_105 = const()[name = tensor<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])];
-            tensor<int32, []> var_106 = const()[name = tensor<string, []>("op_106"), val = tensor<int32, []>(512)];
-            tensor<int32, [1]> var_108 = sub(x = var_106, y = full_sequence_length)[name = tensor<string, []>("op_108")];
-            tensor<bool, [512]> var_109 = less(x = var_105, y = var_108)[name = tensor<string, []>("op_109")];
             tensor<int32, [1]> expand_dims_0_axes_0 = const()[name = tensor<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 = tensor<string, []>("expand_dims_0")];
-            tensor<int32, [2]> var_109_after_broadcast_reps_0 = const()[name = tensor<string, []>("op_109_after_broadcast_reps_0"), val = tensor<int32, [2]>([64, 1])];
-            tensor<bool, [64, 512]> var_109_after_broadcast = tile(reps = var_109_after_broadcast_reps_0, x = expand_dims_0)[name = tensor<string, []>("op_109_after_broadcast")];
-            tensor<fp16, [64, 512]> m_cast_fp16 = select(a = all_mask_to_fp16, b = m_3_cast_fp16, cond = var_109_after_broadcast)[name = tensor<string, []>("m_cast_fp16")];
-            tensor<int32, [1]> var_112_axes_0 = const()[name = tensor<string, []>("op_112_axes_0"), val = tensor<int32, [1]>([0])];
-            tensor<fp16, [1, 64, 512]> var_112_cast_fp16 = expand_dims(axes = var_112_axes_0, x = m_cast_fp16)[name = tensor<string, []>("op_112_cast_fp16")];
-            tensor<int32, [1]> var_114_axes_0 = const()[name = tensor<string, []>("op_114_axes_0"), val = tensor<int32, [1]>([0])];
-            tensor<fp16, [1, 1, 64, 512]> mask = expand_dims(axes = var_114_axes_0, x = var_112_cast_fp16)[name = tensor<string, []>("op_114_cast_fp16")];
         } -> (x, cos, sin, mask);
 }

 program(1.0)
+[buildInfo = dict<tensor<string, []>, tensor<string, []>>({{"coremlc-component-MIL", "3304.5.2"}, {"coremlc-version", "3304.6.2"}, {"coremltools-component-torch", "2.1.0"}, {"coremltools-source-dialect", "TorchScript"}, {"coremltools-version", "8.0b1"}})]
 {
     func main<ios16>(tensor<int32, [1]> full_sequence_length, tensor<int32, [1, 64]> input_ids) {
             tensor<int32, [1]> T = const()[name = tensor<string, []>("T"), val = tensor<int32, [1]>([64])];
+            tensor<int32, []> x_1_axis_0 = const()[name = tensor<string, []>("x_1_axis_0"), val = tensor<int32, []>(0)];
+            tensor<int32, []> x_1_batch_dims_0 = const()[name = tensor<string, []>("x_1_batch_dims_0"), val = tensor<int32, []>(0)];
             tensor<fp16, [32000, 4096]> wte_weight_to_fp16 = const()[name = tensor<string, []>("wte_weight_to_fp16"), val = tensor<fp16, [32000, 4096]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(64)))];
+            tensor<fp16, [1, 64, 4096]> x_1_cast_fp16 = gather(axis = x_1_axis_0, batch_dims = x_1_batch_dims_0, indices = input_ids, x = wte_weight_to_fp16)[name = tensor<string, []>("x_1_cast_fp16")];
+            tensor<int32, [3]> x_perm_0 = const()[name = tensor<string, []>("x_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
+            tensor<int32, [4]> var_27 = const()[name = tensor<string, []>("op_27"), val = tensor<int32, [4]>([1, 4096, -1, 8])];
+            tensor<fp16, [1, 4096, 64]> x_cast_fp16 = transpose(perm = x_perm_0, x = x_1_cast_fp16)[name = tensor<string, []>("transpose_1")];
+            tensor<fp16, [1, 4096, 8, 8]> x = reshape(shape = var_27, x = x_cast_fp16)[name = tensor<string, []>("op_28_cast_fp16")];
             tensor<int32, [1]> pos_offset = sub(x = T, y = full_sequence_length)[name = tensor<string, []>("pos_offset")];
+            tensor<int32, [64]> var_36 = const()[name = tensor<string, []>("op_36"), val = tensor<int32, [64]>([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])];
+            tensor<int32, [64]> input_pos_1 = sub(x = var_36, y = pos_offset)[name = tensor<string, []>("input_pos_1")];
+            tensor<int32, [64]> var_44 = const()[name = tensor<string, []>("op_44"), val = tensor<int32, [64]>([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 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, [64]> input_pos = maximum(x = input_pos_1, y = var_44)[name = tensor<string, []>("input_pos")];
+            tensor<int32, []> var_55 = const()[name = tensor<string, []>("op_55"), val = tensor<int32, []>(1)];
+            tensor<int32, []> var_56_batch_dims_0 = const()[name = tensor<string, []>("op_56_batch_dims_0"), val = tensor<int32, []>(0)];
+            tensor<fp16, [128, 512]> var_54_to_fp16 = const()[name = tensor<string, []>("op_54_to_fp16"), val = tensor<fp16, [128, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(262144128)))];
+            tensor<fp16, [128, 64]> cos = gather(axis = var_55, batch_dims = var_56_batch_dims_0, indices = input_pos, x = var_54_to_fp16)[name = tensor<string, []>("op_56_cast_fp16")];
+            tensor<int32, []> var_66 = const()[name = tensor<string, []>("op_66"), val = tensor<int32, []>(1)];
+            tensor<int32, []> var_67_batch_dims_0 = const()[name = tensor<string, []>("op_67_batch_dims_0"), val = tensor<int32, []>(0)];
+            tensor<fp16, [128, 512]> var_65_to_fp16 = const()[name = tensor<string, []>("op_65_to_fp16"), val = tensor<fp16, [128, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(262275264)))];
+            tensor<fp16, [128, 64]> sin = gather(axis = var_66, batch_dims = var_67_batch_dims_0, indices = input_pos, x = var_65_to_fp16)[name = tensor<string, []>("op_67_cast_fp16")];
+            tensor<int32, [64, 1]> var_102 = const()[name = tensor<string, []>("op_102"), val = tensor<int32, [64, 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]])];
+            tensor<bool, [64, 1]> var_105 = less(x = var_102, y = pos_offset)[name = tensor<string, []>("op_105")];
+            tensor<int32, [2]> var_105_after_broadcast_reps_0 = const()[name = tensor<string, []>("op_105_after_broadcast_reps_0"), val = tensor<int32, [2]>([1, 512])];
+            tensor<bool, [64, 512]> var_105_after_broadcast = tile(reps = var_105_after_broadcast_reps_0, x = var_105)[name = tensor<string, []>("op_105_after_broadcast")];
             tensor<fp16, [64, 512]> all_mask_to_fp16 = const()[name = tensor<string, []>("all_mask_to_fp16"), val = tensor<fp16, [64, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(262406400)))];
             tensor<fp16, [64, 512]> m_1_to_fp16 = const()[name = tensor<string, []>("m_1_to_fp16"), val = tensor<fp16, [64, 512]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(262472000)))];
+            tensor<fp16, [64, 512]> m_3_cast_fp16 = select(a = all_mask_to_fp16, b = m_1_to_fp16, cond = var_105_after_broadcast)[name = tensor<string, []>("m_3_cast_fp16")];
+            tensor<int32, [512]> var_115 = const()[name = tensor<string, []>("op_115"), 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, []> var_116 = const()[name = tensor<string, []>("op_116"), val = tensor<int32, []>(512)];
+            tensor<int32, [1]> var_118 = sub(x = var_116, y = full_sequence_length)[name = tensor<string, []>("op_118")];
+            tensor<bool, [512]> var_119 = less(x = var_115, y = var_118)[name = tensor<string, []>("op_119")];
             tensor<int32, [1]> expand_dims_0_axes_0 = const()[name = tensor<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_119)[name = tensor<string, []>("expand_dims_0")];
+            tensor<int32, [2]> var_119_after_broadcast_reps_0 = const()[name = tensor<string, []>("op_119_after_broadcast_reps_0"), val = tensor<int32, [2]>([64, 1])];
+            tensor<bool, [64, 512]> var_119_after_broadcast = tile(reps = var_119_after_broadcast_reps_0, x = expand_dims_0)[name = tensor<string, []>("op_119_after_broadcast")];
+            tensor<fp16, [64, 512]> m_cast_fp16 = select(a = all_mask_to_fp16, b = m_3_cast_fp16, cond = var_119_after_broadcast)[name = tensor<string, []>("m_cast_fp16")];
+            tensor<int32, [1]> var_122_axes_0 = const()[name = tensor<string, []>("op_122_axes_0"), val = tensor<int32, [1]>([0])];
+            tensor<fp16, [1, 64, 512]> var_122_cast_fp16 = expand_dims(axes = var_122_axes_0, x = m_cast_fp16)[name = tensor<string, []>("op_122_cast_fp16")];
+            tensor<int32, [1]> var_124_axes_0 = const()[name = tensor<string, []>("op_124_axes_0"), val = tensor<int32, [1]>([0])];
+            tensor<fp16, [1, 1, 64, 512]> var_124_cast_fp16 = expand_dims(axes = var_124_axes_0, x = var_122_cast_fp16)[name = tensor<string, []>("op_124_cast_fp16")];
+            tensor<int32, [4]> var_129 = const()[name = tensor<string, []>("op_129"), val = tensor<int32, [4]>([0, 3, 1, 2])];
+            tensor<fp16, [1, 512, 1, 64]> mask = transpose(perm = var_129, x = var_124_cast_fp16)[name = tensor<string, []>("transpose_0")];
         } -> (x, cos, sin, mask);
 }

Llama-2-7b-hf_chunk1.mlmodelc/weights/weight.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:75a8ba0e4d6fc824f820051588b446e6b72dfb09497a058e443ab071d9b3cbc7
 size 262537600

 version https://git-lfs.github.com/spec/v1
+oid sha256:098ec04b28e4314c7c3e9a3bf0f5ee4e7e9211d49ed367953c5d0fbc0f36d13c
 size 262537600

Llama-2-7b-hf_chunk10.mlmodelc/analytics/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3412284b024b899a736cd77112d4b1a4a5faa19d954259e925ef429f58bd886b
 size 243

 version https://git-lfs.github.com/spec/v1
+oid sha256:997c2b09d10cc368b341f867b52aac1e9325853550f47133cc48a353128e881a
 size 243

Llama-2-7b-hf_chunk10.mlmodelc/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b79e263bb20b8a02d650dad2c3eee71ff787829f337aedacb6cd4e1b61c1ce23
-size 791

 version https://git-lfs.github.com/spec/v1
+oid sha256:3fdd5ca1ab176b28ed33e53920cb3ef99dac8b0e220af01536a3969d5d83f1a5
+size 793

Llama-2-7b-hf_chunk10.mlmodelc/metadata.json CHANGED Viewed

@@ -7,9 +7,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_x",
         "type" : "MultiArray"
       },
@@ -17,9 +17,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_0",
         "type" : "MultiArray"
       },
@@ -27,9 +27,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_1",
         "type" : "MultiArray"
       },
@@ -37,9 +37,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_2",
         "type" : "MultiArray"
       },
@@ -47,9 +47,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_0",
         "type" : "MultiArray"
       },
@@ -57,9 +57,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_1",
         "type" : "MultiArray"
       },
@@ -67,9 +67,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_2",
         "type" : "MultiArray"
       }
@@ -79,17 +79,18 @@
     ],
     "specificationVersion" : 7,
     "mlProgramOperationTypeHistogram" : {
-      "Concat" : 18,
-      "Ios16.rsqrt" : 6,
-      "Ios16.mul" : 63,
-      "SliceByIndex" : 12,
       "Ios16.constexprLutToDense" : 21,
       "Ios16.conv" : 21,
-      "Ios16.add" : 21,
-      "Ios16.reduceMean" : 6,
-      "Ios16.matmul" : 6,
-      "Ios16.softmax" : 3,
-      "Ios16.reshape" : 12,
       "Ios16.silu" : 3
     },
     "computePrecision" : "Mixed (Float16, Int32)",
@@ -108,16 +109,16 @@
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
       "com.github.apple.coremltools.source" : "torch==2.1.0",
-      "com.github.apple.coremltools.version" : "7.2"
     },
     "inputSchema" : [
       {
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 4096, 1, 64]",
         "name" : "x",
         "type" : "MultiArray"
       },
@@ -145,9 +146,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 1 × 64 × 512)",
         "shortDescription" : "",
-        "shape" : "[1, 1, 64, 512]",
         "name" : "mask",
         "type" : "MultiArray"
       },
@@ -155,9 +156,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_0",
         "type" : "MultiArray"
       },
@@ -165,9 +166,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_0",
         "type" : "MultiArray"
       },
@@ -175,9 +176,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_1",
         "type" : "MultiArray"
       },
@@ -185,9 +186,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_1",
         "type" : "MultiArray"
       },
@@ -195,9 +196,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_2",
         "type" : "MultiArray"
       },
@@ -205,14 +206,14 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_2",
         "type" : "MultiArray"
       }
     ],
-    "generatedClassName" : "Llama_2_7b_hf_2024_05_25_14_03_55_chunk10",
     "method" : "predict"
   }
 ]

         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 8 × 8)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 8, 8]",
         "name" : "new_x",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_2",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_2",
         "type" : "MultiArray"
       }
     ],
     "specificationVersion" : 7,
     "mlProgramOperationTypeHistogram" : {
+      "Concat" : 21,
+      "Ios16.mul" : 150,
+      "SliceByIndex" : 300,
       "Ios16.constexprLutToDense" : 21,
+      "Transpose" : 3,
+      "Ios16.einsum" : 192,
       "Ios16.conv" : 21,
+      "Ios16.add" : 108,
+      "Ios16.realDiv" : 6,
+      "Ios16.softmax" : 96,
+      "Ios16.reduceL2Norm" : 6,
+      "Ios16.reshape" : 21,
       "Ios16.silu" : 3
     },
     "computePrecision" : "Mixed (Float16, Int32)",
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
       "com.github.apple.coremltools.source" : "torch==2.1.0",
+      "com.github.apple.coremltools.version" : "8.0b1"
     },
     "inputSchema" : [
       {
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 8 × 8)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 8, 8]",
         "name" : "x",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 512 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 512, 1, 64]",
         "name" : "mask",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_2",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_2",
         "type" : "MultiArray"
       }
     ],
+    "generatedClassName" : "Llama_2_7b_hf_2024_08_09_09_54_41_chunk10",
     "method" : "predict"
   }
 ]

Llama-2-7b-hf_chunk10.mlmodelc/model.mil CHANGED Viewed

The diff for this file is too large to render. See raw diff

Llama-2-7b-hf_chunk10.mlmodelc/weights/weight.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:86d4446017950797cf7896941f17c78be0e7c925911e4555f70b1133d20f77b9
-size 303872704

 version https://git-lfs.github.com/spec/v1
+oid sha256:d316f3ae860ed5659a53ae6c81a8e8a352eaa73244c08f3fdbbfe4fcdd8f1be2
+size 303873856

Llama-2-7b-hf_chunk11.mlmodelc/analytics/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3412284b024b899a736cd77112d4b1a4a5faa19d954259e925ef429f58bd886b
 size 243

 version https://git-lfs.github.com/spec/v1
+oid sha256:997c2b09d10cc368b341f867b52aac1e9325853550f47133cc48a353128e881a
 size 243

Llama-2-7b-hf_chunk11.mlmodelc/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:589729b2995d8ca8246bbb5d92b910207bab816ad67282b0a285bcd2de77f80e
-size 791

 version https://git-lfs.github.com/spec/v1
+oid sha256:3fdd5ca1ab176b28ed33e53920cb3ef99dac8b0e220af01536a3969d5d83f1a5
+size 793

Llama-2-7b-hf_chunk11.mlmodelc/metadata.json CHANGED Viewed

@@ -7,9 +7,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_x",
         "type" : "MultiArray"
       },
@@ -17,9 +17,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_0",
         "type" : "MultiArray"
       },
@@ -27,9 +27,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_1",
         "type" : "MultiArray"
       },
@@ -37,9 +37,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_2",
         "type" : "MultiArray"
       },
@@ -47,9 +47,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_0",
         "type" : "MultiArray"
       },
@@ -57,9 +57,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_1",
         "type" : "MultiArray"
       },
@@ -67,9 +67,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_2",
         "type" : "MultiArray"
       }
@@ -79,17 +79,18 @@
     ],
     "specificationVersion" : 7,
     "mlProgramOperationTypeHistogram" : {
-      "Concat" : 18,
-      "Ios16.rsqrt" : 6,
-      "Ios16.mul" : 63,
-      "SliceByIndex" : 12,
       "Ios16.constexprLutToDense" : 21,
       "Ios16.conv" : 21,
-      "Ios16.add" : 21,
-      "Ios16.reduceMean" : 6,
-      "Ios16.matmul" : 6,
-      "Ios16.softmax" : 3,
-      "Ios16.reshape" : 12,
       "Ios16.silu" : 3
     },
     "computePrecision" : "Mixed (Float16, Int32)",
@@ -108,16 +109,16 @@
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
       "com.github.apple.coremltools.source" : "torch==2.1.0",
-      "com.github.apple.coremltools.version" : "7.2"
     },
     "inputSchema" : [
       {
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 4096, 1, 64]",
         "name" : "x",
         "type" : "MultiArray"
       },
@@ -145,9 +146,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 1 × 64 × 512)",
         "shortDescription" : "",
-        "shape" : "[1, 1, 64, 512]",
         "name" : "mask",
         "type" : "MultiArray"
       },
@@ -155,9 +156,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_0",
         "type" : "MultiArray"
       },
@@ -165,9 +166,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_0",
         "type" : "MultiArray"
       },
@@ -175,9 +176,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_1",
         "type" : "MultiArray"
       },
@@ -185,9 +186,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_1",
         "type" : "MultiArray"
       },
@@ -195,9 +196,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_2",
         "type" : "MultiArray"
       },
@@ -205,14 +206,14 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_2",
         "type" : "MultiArray"
       }
     ],
-    "generatedClassName" : "Llama_2_7b_hf_2024_05_25_14_03_55_chunk11",
     "method" : "predict"
   }
 ]

         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 8 × 8)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 8, 8]",
         "name" : "new_x",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_2",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_2",
         "type" : "MultiArray"
       }
     ],
     "specificationVersion" : 7,
     "mlProgramOperationTypeHistogram" : {
+      "Concat" : 21,
+      "Ios16.mul" : 150,
+      "SliceByIndex" : 300,
       "Ios16.constexprLutToDense" : 21,
+      "Transpose" : 3,
+      "Ios16.einsum" : 192,
       "Ios16.conv" : 21,
+      "Ios16.add" : 108,
+      "Ios16.realDiv" : 6,
+      "Ios16.softmax" : 96,
+      "Ios16.reduceL2Norm" : 6,
+      "Ios16.reshape" : 21,
       "Ios16.silu" : 3
     },
     "computePrecision" : "Mixed (Float16, Int32)",
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
       "com.github.apple.coremltools.source" : "torch==2.1.0",
+      "com.github.apple.coremltools.version" : "8.0b1"
     },
     "inputSchema" : [
       {
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 8 × 8)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 8, 8]",
         "name" : "x",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 512 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 512, 1, 64]",
         "name" : "mask",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_2",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_2",
         "type" : "MultiArray"
       }
     ],
+    "generatedClassName" : "Llama_2_7b_hf_2024_08_09_09_54_41_chunk11",
     "method" : "predict"
   }
 ]

Llama-2-7b-hf_chunk11.mlmodelc/model.mil CHANGED Viewed

The diff for this file is too large to render. See raw diff

Llama-2-7b-hf_chunk11.mlmodelc/weights/weight.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9506f3438a1c857418b2dd28a4631b401f24e3bd606f0427c7adbf510af1e2dc
-size 303872704

 version https://git-lfs.github.com/spec/v1
+oid sha256:baecb2f13f8f1adf620bb882d56cf3b6adbc1fe32d9d07645fb8bdfe81e55d9d
+size 303873856

Llama-2-7b-hf_chunk12.mlmodelc/analytics/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a73e9cc1e9aaa1351af7ee9af6a10c0d8fd805fe2383635cee1714240351b5c2
 size 243

 version https://git-lfs.github.com/spec/v1
+oid sha256:7e2c7d4f699871cef135bb77d11029f37679666a0b22b160dcd45a6ef9be60c9
 size 243

Llama-2-7b-hf_chunk12.mlmodelc/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1e030d81f910b53587cf130f1dba0c1d731ab715ebd6ca0b4f475da21707b21e
-size 651

 version https://git-lfs.github.com/spec/v1
+oid sha256:8148a192361b8ecec063ca1450af657313ff478d464c1737f0a57502021593a7
+size 653

Llama-2-7b-hf_chunk12.mlmodelc/metadata.json CHANGED Viewed

@@ -7,9 +7,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_x",
         "type" : "MultiArray"
       },
@@ -17,9 +17,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_0",
         "type" : "MultiArray"
       },
@@ -27,9 +27,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_1",
         "type" : "MultiArray"
       },
@@ -37,9 +37,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_0",
         "type" : "MultiArray"
       },
@@ -47,9 +47,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_1",
         "type" : "MultiArray"
       }
@@ -59,17 +59,18 @@
     ],
     "specificationVersion" : 7,
     "mlProgramOperationTypeHistogram" : {
-      "Concat" : 12,
-      "Ios16.rsqrt" : 4,
-      "Ios16.mul" : 42,
-      "SliceByIndex" : 8,
       "Ios16.constexprLutToDense" : 14,
       "Ios16.conv" : 14,
-      "Ios16.add" : 14,
-      "Ios16.reduceMean" : 4,
-      "Ios16.matmul" : 4,
-      "Ios16.softmax" : 2,
-      "Ios16.reshape" : 8,
       "Ios16.silu" : 2
     },
     "computePrecision" : "Mixed (Float16, Int32)",
@@ -87,17 +88,17 @@
     },
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
-      "com.github.apple.coremltools.source" : "torch==2.1.0",
-      "com.github.apple.coremltools.version" : "7.2"
     },
     "inputSchema" : [
       {
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 4096, 1, 64]",
         "name" : "x",
         "type" : "MultiArray"
       },
@@ -125,9 +126,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 1 × 64 × 512)",
         "shortDescription" : "",
-        "shape" : "[1, 1, 64, 512]",
         "name" : "mask",
         "type" : "MultiArray"
       },
@@ -135,9 +136,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_0",
         "type" : "MultiArray"
       },
@@ -145,9 +146,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_0",
         "type" : "MultiArray"
       },
@@ -155,9 +156,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_1",
         "type" : "MultiArray"
       },
@@ -165,14 +166,14 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_1",
         "type" : "MultiArray"
       }
     ],
-    "generatedClassName" : "Llama_2_7b_hf_2024_05_25_14_03_55_chunk12",
     "method" : "predict"
   }
 ]

         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 8 × 8)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 8, 8]",
         "name" : "new_x",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_1",
         "type" : "MultiArray"
       }
     ],
     "specificationVersion" : 7,
     "mlProgramOperationTypeHistogram" : {
+      "Concat" : 14,
+      "Ios16.mul" : 100,
+      "SliceByIndex" : 200,
       "Ios16.constexprLutToDense" : 14,
+      "Transpose" : 2,
+      "Ios16.einsum" : 128,
       "Ios16.conv" : 14,
+      "Ios16.add" : 72,
+      "Ios16.realDiv" : 4,
+      "Ios16.softmax" : 64,
+      "Ios16.reduceL2Norm" : 4,
+      "Ios16.reshape" : 14,
       "Ios16.silu" : 2
     },
     "computePrecision" : "Mixed (Float16, Int32)",
     },
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
+      "com.github.apple.coremltools.version" : "8.0b1",
+      "com.github.apple.coremltools.source" : "torch==2.1.0"
     },
     "inputSchema" : [
       {
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 8 × 8)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 8, 8]",
         "name" : "x",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 512 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 512, 1, 64]",
         "name" : "mask",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_1",
         "type" : "MultiArray"
       }
     ],
+    "generatedClassName" : "Llama_2_7b_hf_2024_08_09_09_54_41_chunk12",
     "method" : "predict"
   }
 ]

Llama-2-7b-hf_chunk12.mlmodelc/model.mil CHANGED Viewed

The diff for this file is too large to render. See raw diff

Llama-2-7b-hf_chunk12.mlmodelc/weights/weight.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d3d085d837454b4685bcd36331b09a5b0b329f7ef4da1f2dbed101b7ec075630
-size 202581824

 version https://git-lfs.github.com/spec/v1
+oid sha256:8083e76257cb1c212c04f16e970bef147d1f06e4ccdbbe8d9c5f0ffebeed6d34
+size 202582592

Llama-2-7b-hf_chunk13.mlmodelc/analytics/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:55bbf17f4d2567d045baa3ae69337cad81c45f822491151ed7a5b29327f874f6
 size 243

 version https://git-lfs.github.com/spec/v1
+oid sha256:3fe22d6d09bcc30eac1d0d0e786aff8bd12a35552070af919e88c7c9bcac0405
 size 243

Llama-2-7b-hf_chunk13.mlmodelc/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d860ea43d6f8ebbf70594a29be6231ee1d324bdaf2f26417eb82297acb920e17
-size 309

 version https://git-lfs.github.com/spec/v1
+oid sha256:dcd130ddb6d42e6468c1c9ebc7cb2472a42812a6e04950ef5cf75498e5885fcc
+size 311

Llama-2-7b-hf_chunk13.mlmodelc/metadata.json CHANGED Viewed

@@ -19,15 +19,14 @@
     ],
     "specificationVersion" : 7,
     "mlProgramOperationTypeHistogram" : {
-      "Concat" : 1,
-      "Ios16.add" : 1,
-      "Ios16.mul" : 3,
-      "Ios16.rsqrt" : 1,
       "Transpose" : 1,
-      "Ios16.reshape" : 3,
-      "Ios16.reduceMean" : 1,
       "Ios16.matmul" : 2,
-      "Squeeze" : 1
     },
     "computePrecision" : "Mixed (Float16, Int32)",
     "isUpdatable" : "0",
@@ -45,21 +44,21 @@
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
       "com.github.apple.coremltools.source" : "torch==2.1.0",
-      "com.github.apple.coremltools.version" : "7.2"
     },
     "inputSchema" : [
       {
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 4096, 1, 64]",
         "name" : "x",
         "type" : "MultiArray"
       }
     ],
-    "generatedClassName" : "Llama_2_7b_hf_2024_05_25_14_03_55_chunk13",
     "method" : "predict"
   }
 ]

     ],
     "specificationVersion" : 7,
     "mlProgramOperationTypeHistogram" : {
+      "Concat" : 2,
+      "Ios16.mul" : 2,
+      "Squeeze" : 1,
       "Transpose" : 1,
+      "Ios16.reshape" : 4,
       "Ios16.matmul" : 2,
+      "Ios16.realDiv" : 1,
+      "Ios16.reduceL2Norm" : 1
     },
     "computePrecision" : "Mixed (Float16, Int32)",
     "isUpdatable" : "0",
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
       "com.github.apple.coremltools.source" : "torch==2.1.0",
+      "com.github.apple.coremltools.version" : "8.0b1"
     },
     "inputSchema" : [
       {
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 8 × 8)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 8, 8]",
         "name" : "x",
         "type" : "MultiArray"
       }
     ],
+    "generatedClassName" : "Llama_2_7b_hf_2024_08_09_09_54_41_chunk13",
     "method" : "predict"
   }
 ]

Llama-2-7b-hf_chunk13.mlmodelc/model.mil CHANGED Viewed

@@ -1,38 +1,41 @@
 program(1.0)
-[buildInfo = dict<tensor<string, []>, tensor<string, []>>({{"coremlc-component-MIL", "5.33.5"}, {"coremlc-version", "1877.40.3"}, {"coremltools-component-torch", "2.1.0"}, {"coremltools-source-dialect", "TorchScript"}, {"coremltools-version", "7.2"}})]
 {
-    func main<ios16>(tensor<fp16, [1, 4096, 1, 64]> x) {
             tensor<bool, []> var_6 = const()[name = tensor<string, []>("op_6"), val = tensor<bool, []>(true)];
-            tensor<fp16, [1, 4096, 1, 64]> var_13_cast_fp16 = mul(x = x, y = x)[name = tensor<string, []>("op_13_cast_fp16")];
-            tensor<int32, [1]> var_14 = const()[name = tensor<string, []>("op_14"), val = tensor<int32, [1]>([1])];
-            tensor<fp16, [1, 1, 1, 64]> norm_x_cast_fp16 = reduce_mean(axes = var_14, keep_dims = var_6, x = var_13_cast_fp16)[name = tensor<string, []>("norm_x_cast_fp16")];
-            tensor<fp16, []> var_16_to_fp16 = const()[name = tensor<string, []>("op_16_to_fp16"), val = tensor<fp16, []>(0x1.5p-17)];
-            tensor<fp16, [1, 1, 1, 64]> var_17_cast_fp16 = add(x = norm_x_cast_fp16, y = var_16_to_fp16)[name = tensor<string, []>("op_17_cast_fp16")];
-            tensor<fp16, []> var_18_epsilon_0_to_fp16 = const()[name = tensor<string, []>("op_18_epsilon_0_to_fp16"), val = tensor<fp16, []>(0x1p-24)];
-            tensor<fp16, [1, 1, 1, 64]> var_18_cast_fp16 = rsqrt(epsilon = var_18_epsilon_0_to_fp16, x = var_17_cast_fp16)[name = tensor<string, []>("op_18_cast_fp16")];
-            tensor<fp16, [1, 4096, 1, 64]> x_normed_1_cast_fp16 = mul(x = x, y = var_18_cast_fp16)[name = tensor<string, []>("x_normed_1_cast_fp16")];
-            tensor<fp16, [1, 4096, 1, 1]> ln_f_weight_to_fp16 = const()[name = tensor<string, []>("ln_f_weight_to_fp16"), val = tensor<fp16, [1, 4096, 1, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(64)))];
-            tensor<fp16, [1, 4096, 1, 64]> x_cast_fp16 = mul(x = x_normed_1_cast_fp16, y = ln_f_weight_to_fp16)[name = tensor<string, []>("x_cast_fp16")];
-            tensor<int32, [1]> var_23_axes_0 = const()[name = tensor<string, []>("op_23_axes_0"), val = tensor<int32, [1]>([2])];
-            tensor<fp16, [1, 4096, 64]> var_23_cast_fp16 = squeeze(axes = var_23_axes_0, x = x_cast_fp16)[name = tensor<string, []>("op_23_cast_fp16")];
-            tensor<int32, [3]> var_26_perm_0 = const()[name = tensor<string, []>("op_26_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
             tensor<int32, [2]> concat_4 = const()[name = tensor<string, []>("concat_4"), val = tensor<int32, [2]>([64, 4096])];
-            tensor<fp16, [1, 64, 4096]> transpose_4 = transpose(perm = var_26_perm_0, x = var_23_cast_fp16)[name = tensor<string, []>("transpose_4")];
-            tensor<fp16, [64, 4096]> reshape_0_cast_fp16 = reshape(shape = concat_4, x = transpose_4)[name = tensor<string, []>("reshape_0_cast_fp16")];
             tensor<bool, []> matmul_0_transpose_x_0 = const()[name = tensor<string, []>("matmul_0_transpose_x_0"), val = tensor<bool, []>(false)];
             tensor<bool, []> matmul_0_transpose_y_0 = const()[name = tensor<string, []>("matmul_0_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<fp16, [4096, 16384]> transpose_1_to_fp16 = const()[name = tensor<string, []>("transpose_1_to_fp16"), val = tensor<fp16, [4096, 16384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(8320)))];
             tensor<fp16, [64, 16384]> matmul_0_cast_fp16 = matmul(transpose_x = matmul_0_transpose_x_0, transpose_y = matmul_0_transpose_y_0, x = reshape_0_cast_fp16, y = transpose_1_to_fp16)[name = tensor<string, []>("matmul_0_cast_fp16")];
             tensor<int32, [3]> concat_8 = const()[name = tensor<string, []>("concat_8"), val = tensor<int32, [3]>([1, 64, 16384])];
             tensor<fp16, [1, 64, 16384]> reshape_2_cast_fp16 = reshape(shape = concat_8, x = matmul_0_cast_fp16)[name = tensor<string, []>("reshape_2_cast_fp16")];
             tensor<bool, []> matmul_1_transpose_x_0 = const()[name = tensor<string, []>("matmul_1_transpose_x_0"), val = tensor<bool, []>(false)];
             tensor<bool, []> matmul_1_transpose_y_0 = const()[name = tensor<string, []>("matmul_1_transpose_y_0"), val = tensor<bool, []>(false)];
-            tensor<fp16, [4096, 15616]> transpose_3_to_fp16 = const()[name = tensor<string, []>("transpose_3_to_fp16"), val = tensor<fp16, [4096, 15616]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(134226112)))];
             tensor<fp16, [64, 15616]> matmul_1_cast_fp16 = matmul(transpose_x = matmul_1_transpose_x_0, transpose_y = matmul_1_transpose_y_0, x = reshape_0_cast_fp16, y = transpose_3_to_fp16)[name = tensor<string, []>("matmul_1_cast_fp16")];
             tensor<int32, [3]> concat_16 = const()[name = tensor<string, []>("concat_16"), val = tensor<int32, [3]>([1, 64, 15616])];
             tensor<fp16, [1, 64, 15616]> reshape_5_cast_fp16 = reshape(shape = concat_16, x = matmul_1_cast_fp16)[name = tensor<string, []>("reshape_5_cast_fp16")];
-            tensor<int32, []> var_41 = const()[name = tensor<string, []>("op_41"), val = tensor<int32, []>(-1)];
-            tensor<bool, []> var_42_interleave_0 = const()[name = tensor<string, []>("op_42_interleave_0"), val = tensor<bool, []>(false)];
-            tensor<fp16, [1, 64, 32000]> logits = concat(axis = var_41, interleave = var_42_interleave_0, values = (reshape_2_cast_fp16, reshape_5_cast_fp16))[name = tensor<string, []>("op_42_cast_fp16")];
         } -> (logits);
 }

 program(1.0)
+[buildInfo = dict<tensor<string, []>, tensor<string, []>>({{"coremlc-component-MIL", "3304.5.2"}, {"coremlc-version", "3304.6.2"}, {"coremltools-component-torch", "2.1.0"}, {"coremltools-source-dialect", "TorchScript"}, {"coremltools-version", "8.0b1"}})]
 {
+    func main<ios16>(tensor<fp16, [1, 4096, 8, 8]> x) {
             tensor<bool, []> var_6 = const()[name = tensor<string, []>("op_6"), val = tensor<bool, []>(true)];
+            tensor<int32, []> var_9 = const()[name = tensor<string, []>("op_9"), val = tensor<int32, []>(1)];
+            tensor<bool, []> x_eps_interleave_0 = const()[name = tensor<string, []>("x_eps_interleave_0"), val = tensor<bool, []>(false)];
+            tensor<fp16, [1, 1, 8, 8]> eps_chan_to_fp16 = const()[name = tensor<string, []>("eps_chan_to_fp16"), val = tensor<fp16, [1, 1, 8, 8]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(64)))];
+            tensor<fp16, [1, 4097, 8, 8]> x_eps_cast_fp16 = concat(axis = var_9, interleave = x_eps_interleave_0, values = (x, eps_chan_to_fp16))[name = tensor<string, []>("x_eps_cast_fp16")];
+            tensor<int32, [1]> norm_x_axes_0 = const()[name = tensor<string, []>("norm_x_axes_0"), val = tensor<int32, [1]>([1])];
+            tensor<fp16, [1, 1, 8, 8]> norm_x_cast_fp16 = reduce_l2_norm(axes = norm_x_axes_0, keep_dims = var_6, x = x_eps_cast_fp16)[name = tensor<string, []>("norm_x_cast_fp16")];
+            tensor<fp16, [1, 4096, 8, 8]> x_normed_1_cast_fp16 = real_div(x = x, y = norm_x_cast_fp16)[name = tensor<string, []>("x_normed_1_cast_fp16")];
+            tensor<fp16, []> var_34_to_fp16 = const()[name = tensor<string, []>("op_34_to_fp16"), val = tensor<fp16, []>(0x1p+6)];
+            tensor<fp16, [1, 4096, 8, 8]> x_normed_3_cast_fp16 = mul(x = x_normed_1_cast_fp16, y = var_34_to_fp16)[name = tensor<string, []>("x_normed_3_cast_fp16")];
+            tensor<fp16, [1, 4096, 1, 1]> ln_f_weight_to_fp16 = const()[name = tensor<string, []>("ln_f_weight_to_fp16"), val = tensor<fp16, [1, 4096, 1, 1]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(256)))];
+            tensor<fp16, [1, 4096, 8, 8]> x_5_cast_fp16 = mul(x = x_normed_3_cast_fp16, y = ln_f_weight_to_fp16)[name = tensor<string, []>("x_5_cast_fp16")];
+            tensor<int32, [4]> var_48 = const()[name = tensor<string, []>("op_48"), val = tensor<int32, [4]>([1, 4096, 1, -1])];
+            tensor<fp16, [1, 4096, 1, 64]> x_cast_fp16 = reshape(shape = var_48, x = x_5_cast_fp16)[name = tensor<string, []>("x_cast_fp16")];
+            tensor<int32, [1]> var_51_axes_0 = const()[name = tensor<string, []>("op_51_axes_0"), val = tensor<int32, [1]>([2])];
+            tensor<fp16, [1, 4096, 64]> var_51_cast_fp16 = squeeze(axes = var_51_axes_0, x = x_cast_fp16)[name = tensor<string, []>("op_51_cast_fp16")];
+            tensor<int32, [3]> var_54_perm_0 = const()[name = tensor<string, []>("op_54_perm_0"), val = tensor<int32, [3]>([0, 2, 1])];
             tensor<int32, [2]> concat_4 = const()[name = tensor<string, []>("concat_4"), val = tensor<int32, [2]>([64, 4096])];
+            tensor<fp16, [1, 64, 4096]> var_54_cast_fp16 = transpose(perm = var_54_perm_0, x = var_51_cast_fp16)[name = tensor<string, []>("transpose_4")];
+            tensor<fp16, [64, 4096]> reshape_0_cast_fp16 = reshape(shape = concat_4, x = var_54_cast_fp16)[name = tensor<string, []>("reshape_0_cast_fp16")];
             tensor<bool, []> matmul_0_transpose_x_0 = const()[name = tensor<string, []>("matmul_0_transpose_x_0"), val = tensor<bool, []>(false)];
             tensor<bool, []> matmul_0_transpose_y_0 = const()[name = tensor<string, []>("matmul_0_transpose_y_0"), val = tensor<bool, []>(false)];
+            tensor<fp16, [4096, 16384]> transpose_1_to_fp16 = const()[name = tensor<string, []>("transpose_1_to_fp16"), val = tensor<fp16, [4096, 16384]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(8512)))];
             tensor<fp16, [64, 16384]> matmul_0_cast_fp16 = matmul(transpose_x = matmul_0_transpose_x_0, transpose_y = matmul_0_transpose_y_0, x = reshape_0_cast_fp16, y = transpose_1_to_fp16)[name = tensor<string, []>("matmul_0_cast_fp16")];
             tensor<int32, [3]> concat_8 = const()[name = tensor<string, []>("concat_8"), val = tensor<int32, [3]>([1, 64, 16384])];
             tensor<fp16, [1, 64, 16384]> reshape_2_cast_fp16 = reshape(shape = concat_8, x = matmul_0_cast_fp16)[name = tensor<string, []>("reshape_2_cast_fp16")];
             tensor<bool, []> matmul_1_transpose_x_0 = const()[name = tensor<string, []>("matmul_1_transpose_x_0"), val = tensor<bool, []>(false)];
             tensor<bool, []> matmul_1_transpose_y_0 = const()[name = tensor<string, []>("matmul_1_transpose_y_0"), val = tensor<bool, []>(false)];
+            tensor<fp16, [4096, 15616]> transpose_3_to_fp16 = const()[name = tensor<string, []>("transpose_3_to_fp16"), val = tensor<fp16, [4096, 15616]>(BLOBFILE(path = tensor<string, []>("@model_path/weights/weight.bin"), offset = tensor<uint64, []>(134226304)))];
             tensor<fp16, [64, 15616]> matmul_1_cast_fp16 = matmul(transpose_x = matmul_1_transpose_x_0, transpose_y = matmul_1_transpose_y_0, x = reshape_0_cast_fp16, y = transpose_3_to_fp16)[name = tensor<string, []>("matmul_1_cast_fp16")];
             tensor<int32, [3]> concat_16 = const()[name = tensor<string, []>("concat_16"), val = tensor<int32, [3]>([1, 64, 15616])];
             tensor<fp16, [1, 64, 15616]> reshape_5_cast_fp16 = reshape(shape = concat_16, x = matmul_1_cast_fp16)[name = tensor<string, []>("reshape_5_cast_fp16")];
+            tensor<int32, []> var_69 = const()[name = tensor<string, []>("op_69"), val = tensor<int32, []>(-1)];
+            tensor<bool, []> var_70_interleave_0 = const()[name = tensor<string, []>("op_70_interleave_0"), val = tensor<bool, []>(false)];
+            tensor<fp16, [1, 64, 32000]> logits = concat(axis = var_69, interleave = var_70_interleave_0, values = (reshape_2_cast_fp16, reshape_5_cast_fp16))[name = tensor<string, []>("op_70_cast_fp16")];
         } -> (logits);
 }

Llama-2-7b-hf_chunk13.mlmodelc/weights/weight.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:23cc0c8382a52638c94e9c9963873d35d3222e897233b39b03f4cc92deae2edb
-size 262152448

 version https://git-lfs.github.com/spec/v1
+oid sha256:a041617e42d7a89d6d1e60a90971f24f8fa62634d1c5db56abd302dcf9c3398e
+size 262152640

Llama-2-7b-hf_chunk2.mlmodelc/analytics/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3412284b024b899a736cd77112d4b1a4a5faa19d954259e925ef429f58bd886b
 size 243

 version https://git-lfs.github.com/spec/v1
+oid sha256:997c2b09d10cc368b341f867b52aac1e9325853550f47133cc48a353128e881a
 size 243

Llama-2-7b-hf_chunk2.mlmodelc/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:589729b2995d8ca8246bbb5d92b910207bab816ad67282b0a285bcd2de77f80e
-size 791

 version https://git-lfs.github.com/spec/v1
+oid sha256:ed0dd15fc572d2cc2ec19b317245622b4256a8737cc9ba114529e925d3bf42f2
+size 793

Llama-2-7b-hf_chunk2.mlmodelc/metadata.json CHANGED Viewed

@@ -7,9 +7,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_x",
         "type" : "MultiArray"
       },
@@ -17,9 +17,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_0",
         "type" : "MultiArray"
       },
@@ -27,9 +27,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_1",
         "type" : "MultiArray"
       },
@@ -37,9 +37,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_2",
         "type" : "MultiArray"
       },
@@ -47,9 +47,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_0",
         "type" : "MultiArray"
       },
@@ -57,9 +57,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_1",
         "type" : "MultiArray"
       },
@@ -67,9 +67,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_2",
         "type" : "MultiArray"
       }
@@ -79,17 +79,18 @@
     ],
     "specificationVersion" : 7,
     "mlProgramOperationTypeHistogram" : {
-      "Concat" : 18,
-      "Ios16.rsqrt" : 6,
-      "Ios16.mul" : 63,
-      "SliceByIndex" : 12,
       "Ios16.constexprLutToDense" : 21,
       "Ios16.conv" : 21,
-      "Ios16.add" : 21,
-      "Ios16.reduceMean" : 6,
-      "Ios16.matmul" : 6,
-      "Ios16.softmax" : 3,
-      "Ios16.reshape" : 12,
       "Ios16.silu" : 3
     },
     "computePrecision" : "Mixed (Float16, Int32)",
@@ -108,16 +109,16 @@
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
       "com.github.apple.coremltools.source" : "torch==2.1.0",
-      "com.github.apple.coremltools.version" : "7.2"
     },
     "inputSchema" : [
       {
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 4096, 1, 64]",
         "name" : "x",
         "type" : "MultiArray"
       },
@@ -145,9 +146,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 1 × 64 × 512)",
         "shortDescription" : "",
-        "shape" : "[1, 1, 64, 512]",
         "name" : "mask",
         "type" : "MultiArray"
       },
@@ -155,9 +156,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_0",
         "type" : "MultiArray"
       },
@@ -165,9 +166,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_0",
         "type" : "MultiArray"
       },
@@ -175,9 +176,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_1",
         "type" : "MultiArray"
       },
@@ -185,9 +186,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_1",
         "type" : "MultiArray"
       },
@@ -195,9 +196,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_2",
         "type" : "MultiArray"
       },
@@ -205,14 +206,14 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_2",
         "type" : "MultiArray"
       }
     ],
-    "generatedClassName" : "Llama_2_7b_hf_2024_05_25_14_03_55_chunk2",
     "method" : "predict"
   }
 ]

         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 8 × 8)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 8, 8]",
         "name" : "new_x",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_2",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_2",
         "type" : "MultiArray"
       }
     ],
     "specificationVersion" : 7,
     "mlProgramOperationTypeHistogram" : {
+      "Concat" : 21,
+      "Ios16.mul" : 150,
+      "SliceByIndex" : 300,
       "Ios16.constexprLutToDense" : 21,
+      "Transpose" : 3,
+      "Ios16.einsum" : 192,
       "Ios16.conv" : 21,
+      "Ios16.add" : 108,
+      "Ios16.realDiv" : 6,
+      "Ios16.softmax" : 96,
+      "Ios16.reduceL2Norm" : 6,
+      "Ios16.reshape" : 21,
       "Ios16.silu" : 3
     },
     "computePrecision" : "Mixed (Float16, Int32)",
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
       "com.github.apple.coremltools.source" : "torch==2.1.0",
+      "com.github.apple.coremltools.version" : "8.0b1"
     },
     "inputSchema" : [
       {
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 8 × 8)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 8, 8]",
         "name" : "x",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 512 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 512, 1, 64]",
         "name" : "mask",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_2",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_2",
         "type" : "MultiArray"
       }
     ],
+    "generatedClassName" : "Llama_2_7b_hf_2024_08_09_09_54_41_chunk2",
     "method" : "predict"
   }
 ]

Llama-2-7b-hf_chunk2.mlmodelc/model.mil CHANGED Viewed

The diff for this file is too large to render. See raw diff

Llama-2-7b-hf_chunk2.mlmodelc/weights/weight.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d25138904c91ffd7e03563365ae012b5b126a2b75fc66880152e092e7680e211
-size 303872704

 version https://git-lfs.github.com/spec/v1
+oid sha256:986f04c4d84fa222e29a878e94d10efdb5cb893883201ece3a0a060f9ab5066e
+size 303873856

Llama-2-7b-hf_chunk3.mlmodelc/analytics/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3412284b024b899a736cd77112d4b1a4a5faa19d954259e925ef429f58bd886b
 size 243

 version https://git-lfs.github.com/spec/v1
+oid sha256:997c2b09d10cc368b341f867b52aac1e9325853550f47133cc48a353128e881a
 size 243

Llama-2-7b-hf_chunk3.mlmodelc/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:589729b2995d8ca8246bbb5d92b910207bab816ad67282b0a285bcd2de77f80e
-size 791

 version https://git-lfs.github.com/spec/v1
+oid sha256:3fdd5ca1ab176b28ed33e53920cb3ef99dac8b0e220af01536a3969d5d83f1a5
+size 793

Llama-2-7b-hf_chunk3.mlmodelc/metadata.json CHANGED Viewed

@@ -7,9 +7,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_x",
         "type" : "MultiArray"
       },
@@ -17,9 +17,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_0",
         "type" : "MultiArray"
       },
@@ -27,9 +27,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_1",
         "type" : "MultiArray"
       },
@@ -37,9 +37,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_2",
         "type" : "MultiArray"
       },
@@ -47,9 +47,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_0",
         "type" : "MultiArray"
       },
@@ -57,9 +57,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_1",
         "type" : "MultiArray"
       },
@@ -67,9 +67,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_2",
         "type" : "MultiArray"
       }
@@ -79,17 +79,18 @@
     ],
     "specificationVersion" : 7,
     "mlProgramOperationTypeHistogram" : {
-      "Concat" : 18,
-      "Ios16.rsqrt" : 6,
-      "Ios16.mul" : 63,
-      "SliceByIndex" : 12,
       "Ios16.constexprLutToDense" : 21,
       "Ios16.conv" : 21,
-      "Ios16.add" : 21,
-      "Ios16.reduceMean" : 6,
-      "Ios16.matmul" : 6,
-      "Ios16.softmax" : 3,
-      "Ios16.reshape" : 12,
       "Ios16.silu" : 3
     },
     "computePrecision" : "Mixed (Float16, Int32)",
@@ -108,16 +109,16 @@
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
       "com.github.apple.coremltools.source" : "torch==2.1.0",
-      "com.github.apple.coremltools.version" : "7.2"
     },
     "inputSchema" : [
       {
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 4096, 1, 64]",
         "name" : "x",
         "type" : "MultiArray"
       },
@@ -145,9 +146,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 1 × 64 × 512)",
         "shortDescription" : "",
-        "shape" : "[1, 1, 64, 512]",
         "name" : "mask",
         "type" : "MultiArray"
       },
@@ -155,9 +156,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_0",
         "type" : "MultiArray"
       },
@@ -165,9 +166,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_0",
         "type" : "MultiArray"
       },
@@ -175,9 +176,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_1",
         "type" : "MultiArray"
       },
@@ -185,9 +186,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_1",
         "type" : "MultiArray"
       },
@@ -195,9 +196,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_2",
         "type" : "MultiArray"
       },
@@ -205,14 +206,14 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_2",
         "type" : "MultiArray"
       }
     ],
-    "generatedClassName" : "Llama_2_7b_hf_2024_05_25_14_03_55_chunk3",
     "method" : "predict"
   }
 ]

         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 8 × 8)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 8, 8]",
         "name" : "new_x",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_2",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_2",
         "type" : "MultiArray"
       }
     ],
     "specificationVersion" : 7,
     "mlProgramOperationTypeHistogram" : {
+      "Concat" : 21,
+      "Ios16.mul" : 150,
+      "SliceByIndex" : 300,
       "Ios16.constexprLutToDense" : 21,
+      "Transpose" : 3,
+      "Ios16.einsum" : 192,
       "Ios16.conv" : 21,
+      "Ios16.add" : 108,
+      "Ios16.realDiv" : 6,
+      "Ios16.softmax" : 96,
+      "Ios16.reduceL2Norm" : 6,
+      "Ios16.reshape" : 21,
       "Ios16.silu" : 3
     },
     "computePrecision" : "Mixed (Float16, Int32)",
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
       "com.github.apple.coremltools.source" : "torch==2.1.0",
+      "com.github.apple.coremltools.version" : "8.0b1"
     },
     "inputSchema" : [
       {
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 8 × 8)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 8, 8]",
         "name" : "x",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 512 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 512, 1, 64]",
         "name" : "mask",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_2",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_2",
         "type" : "MultiArray"
       }
     ],
+    "generatedClassName" : "Llama_2_7b_hf_2024_08_09_09_54_41_chunk3",
     "method" : "predict"
   }
 ]

Llama-2-7b-hf_chunk3.mlmodelc/model.mil CHANGED Viewed

The diff for this file is too large to render. See raw diff

Llama-2-7b-hf_chunk3.mlmodelc/weights/weight.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ad1bc13ecfabbb4f02f8306bf18913019826fb28b002e14f11bddeca7a9edefa
-size 303872704

 version https://git-lfs.github.com/spec/v1
+oid sha256:0afd4ddd01084d67eec0c37c977b0ac2be78fe40cf7afcbe707e988e4060ffec
+size 303873856

Llama-2-7b-hf_chunk4.mlmodelc/analytics/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3412284b024b899a736cd77112d4b1a4a5faa19d954259e925ef429f58bd886b
 size 243

 version https://git-lfs.github.com/spec/v1
+oid sha256:997c2b09d10cc368b341f867b52aac1e9325853550f47133cc48a353128e881a
 size 243

Llama-2-7b-hf_chunk4.mlmodelc/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b79e263bb20b8a02d650dad2c3eee71ff787829f337aedacb6cd4e1b61c1ce23
-size 791

 version https://git-lfs.github.com/spec/v1
+oid sha256:3fdd5ca1ab176b28ed33e53920cb3ef99dac8b0e220af01536a3969d5d83f1a5
+size 793

Llama-2-7b-hf_chunk4.mlmodelc/metadata.json CHANGED Viewed

@@ -7,9 +7,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_x",
         "type" : "MultiArray"
       },
@@ -17,9 +17,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_0",
         "type" : "MultiArray"
       },
@@ -27,9 +27,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_1",
         "type" : "MultiArray"
       },
@@ -37,9 +37,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_2",
         "type" : "MultiArray"
       },
@@ -47,9 +47,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_0",
         "type" : "MultiArray"
       },
@@ -57,9 +57,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_1",
         "type" : "MultiArray"
       },
@@ -67,9 +67,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_2",
         "type" : "MultiArray"
       }
@@ -79,17 +79,18 @@
     ],
     "specificationVersion" : 7,
     "mlProgramOperationTypeHistogram" : {
-      "Concat" : 18,
-      "Ios16.rsqrt" : 6,
-      "Ios16.mul" : 63,
-      "SliceByIndex" : 12,
       "Ios16.constexprLutToDense" : 21,
       "Ios16.conv" : 21,
-      "Ios16.add" : 21,
-      "Ios16.reduceMean" : 6,
-      "Ios16.matmul" : 6,
-      "Ios16.softmax" : 3,
-      "Ios16.reshape" : 12,
       "Ios16.silu" : 3
     },
     "computePrecision" : "Mixed (Float16, Int32)",
@@ -108,16 +109,16 @@
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
       "com.github.apple.coremltools.source" : "torch==2.1.0",
-      "com.github.apple.coremltools.version" : "7.2"
     },
     "inputSchema" : [
       {
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 4096, 1, 64]",
         "name" : "x",
         "type" : "MultiArray"
       },
@@ -145,9 +146,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 1 × 64 × 512)",
         "shortDescription" : "",
-        "shape" : "[1, 1, 64, 512]",
         "name" : "mask",
         "type" : "MultiArray"
       },
@@ -155,9 +156,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_0",
         "type" : "MultiArray"
       },
@@ -165,9 +166,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_0",
         "type" : "MultiArray"
       },
@@ -175,9 +176,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_1",
         "type" : "MultiArray"
       },
@@ -185,9 +186,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_1",
         "type" : "MultiArray"
       },
@@ -195,9 +196,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_2",
         "type" : "MultiArray"
       },
@@ -205,14 +206,14 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_2",
         "type" : "MultiArray"
       }
     ],
-    "generatedClassName" : "Llama_2_7b_hf_2024_05_25_14_03_55_chunk4",
     "method" : "predict"
   }
 ]

         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 8 × 8)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 8, 8]",
         "name" : "new_x",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_2",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_2",
         "type" : "MultiArray"
       }
     ],
     "specificationVersion" : 7,
     "mlProgramOperationTypeHistogram" : {
+      "Concat" : 21,
+      "Ios16.mul" : 150,
+      "SliceByIndex" : 300,
       "Ios16.constexprLutToDense" : 21,
+      "Transpose" : 3,
+      "Ios16.einsum" : 192,
       "Ios16.conv" : 21,
+      "Ios16.add" : 108,
+      "Ios16.realDiv" : 6,
+      "Ios16.softmax" : 96,
+      "Ios16.reduceL2Norm" : 6,
+      "Ios16.reshape" : 21,
       "Ios16.silu" : 3
     },
     "computePrecision" : "Mixed (Float16, Int32)",
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
       "com.github.apple.coremltools.source" : "torch==2.1.0",
+      "com.github.apple.coremltools.version" : "8.0b1"
     },
     "inputSchema" : [
       {
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 8 × 8)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 8, 8]",
         "name" : "x",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 512 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 512, 1, 64]",
         "name" : "mask",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_2",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_2",
         "type" : "MultiArray"
       }
     ],
+    "generatedClassName" : "Llama_2_7b_hf_2024_08_09_09_54_41_chunk4",
     "method" : "predict"
   }
 ]

Llama-2-7b-hf_chunk4.mlmodelc/model.mil CHANGED Viewed

The diff for this file is too large to render. See raw diff

Llama-2-7b-hf_chunk4.mlmodelc/weights/weight.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a2b1969a0b2372ca72340108bf7967f643d02a423cac947a5bd3608fdde48b86
-size 303872704

 version https://git-lfs.github.com/spec/v1
+oid sha256:d368880511e37c9082ff760946ff94910ec04d00ebd70f03242588bceb67a685
+size 303873856

Llama-2-7b-hf_chunk5.mlmodelc/analytics/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3412284b024b899a736cd77112d4b1a4a5faa19d954259e925ef429f58bd886b
 size 243

 version https://git-lfs.github.com/spec/v1
+oid sha256:997c2b09d10cc368b341f867b52aac1e9325853550f47133cc48a353128e881a
 size 243

Llama-2-7b-hf_chunk5.mlmodelc/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:589729b2995d8ca8246bbb5d92b910207bab816ad67282b0a285bcd2de77f80e
-size 791

 version https://git-lfs.github.com/spec/v1
+oid sha256:ed0dd15fc572d2cc2ec19b317245622b4256a8737cc9ba114529e925d3bf42f2
+size 793

Llama-2-7b-hf_chunk5.mlmodelc/metadata.json CHANGED Viewed

@@ -7,9 +7,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_x",
         "type" : "MultiArray"
       },
@@ -17,9 +17,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_0",
         "type" : "MultiArray"
       },
@@ -27,9 +27,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_1",
         "type" : "MultiArray"
       },
@@ -37,9 +37,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_2",
         "type" : "MultiArray"
       },
@@ -47,9 +47,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_0",
         "type" : "MultiArray"
       },
@@ -57,9 +57,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_1",
         "type" : "MultiArray"
       },
@@ -67,9 +67,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_2",
         "type" : "MultiArray"
       }
@@ -79,17 +79,18 @@
     ],
     "specificationVersion" : 7,
     "mlProgramOperationTypeHistogram" : {
-      "Concat" : 18,
-      "Ios16.rsqrt" : 6,
-      "Ios16.mul" : 63,
-      "SliceByIndex" : 12,
       "Ios16.constexprLutToDense" : 21,
       "Ios16.conv" : 21,
-      "Ios16.add" : 21,
-      "Ios16.reduceMean" : 6,
-      "Ios16.matmul" : 6,
-      "Ios16.softmax" : 3,
-      "Ios16.reshape" : 12,
       "Ios16.silu" : 3
     },
     "computePrecision" : "Mixed (Float16, Int32)",
@@ -108,16 +109,16 @@
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
       "com.github.apple.coremltools.source" : "torch==2.1.0",
-      "com.github.apple.coremltools.version" : "7.2"
     },
     "inputSchema" : [
       {
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 4096, 1, 64]",
         "name" : "x",
         "type" : "MultiArray"
       },
@@ -145,9 +146,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 1 × 64 × 512)",
         "shortDescription" : "",
-        "shape" : "[1, 1, 64, 512]",
         "name" : "mask",
         "type" : "MultiArray"
       },
@@ -155,9 +156,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_0",
         "type" : "MultiArray"
       },
@@ -165,9 +166,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_0",
         "type" : "MultiArray"
       },
@@ -175,9 +176,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_1",
         "type" : "MultiArray"
       },
@@ -185,9 +186,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_1",
         "type" : "MultiArray"
       },
@@ -195,9 +196,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_2",
         "type" : "MultiArray"
       },
@@ -205,14 +206,14 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_2",
         "type" : "MultiArray"
       }
     ],
-    "generatedClassName" : "Llama_2_7b_hf_2024_05_25_14_03_55_chunk5",
     "method" : "predict"
   }
 ]

         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 8 × 8)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 8, 8]",
         "name" : "new_x",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_2",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_2",
         "type" : "MultiArray"
       }
     ],
     "specificationVersion" : 7,
     "mlProgramOperationTypeHistogram" : {
+      "Concat" : 21,
+      "Ios16.mul" : 150,
+      "SliceByIndex" : 300,
       "Ios16.constexprLutToDense" : 21,
+      "Transpose" : 3,
+      "Ios16.einsum" : 192,
       "Ios16.conv" : 21,
+      "Ios16.add" : 108,
+      "Ios16.realDiv" : 6,
+      "Ios16.softmax" : 96,
+      "Ios16.reduceL2Norm" : 6,
+      "Ios16.reshape" : 21,
       "Ios16.silu" : 3
     },
     "computePrecision" : "Mixed (Float16, Int32)",
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
       "com.github.apple.coremltools.source" : "torch==2.1.0",
+      "com.github.apple.coremltools.version" : "8.0b1"
     },
     "inputSchema" : [
       {
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 8 × 8)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 8, 8]",
         "name" : "x",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 512 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 512, 1, 64]",
         "name" : "mask",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_2",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_2",
         "type" : "MultiArray"
       }
     ],
+    "generatedClassName" : "Llama_2_7b_hf_2024_08_09_09_54_41_chunk5",
     "method" : "predict"
   }
 ]

Llama-2-7b-hf_chunk5.mlmodelc/model.mil CHANGED Viewed

The diff for this file is too large to render. See raw diff

Llama-2-7b-hf_chunk5.mlmodelc/weights/weight.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5d931534284a44e5004b85274be8d122ee55af90a599ea689a9491c6ce13fa16
-size 303872704

 version https://git-lfs.github.com/spec/v1
+oid sha256:d28027700ce1e1dc02bc839d431de13ae2df66b53637a7b559799adce3b84afc
+size 303873856

Llama-2-7b-hf_chunk6.mlmodelc/analytics/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3412284b024b899a736cd77112d4b1a4a5faa19d954259e925ef429f58bd886b
 size 243

 version https://git-lfs.github.com/spec/v1
+oid sha256:997c2b09d10cc368b341f867b52aac1e9325853550f47133cc48a353128e881a
 size 243

Llama-2-7b-hf_chunk6.mlmodelc/coremldata.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:589729b2995d8ca8246bbb5d92b910207bab816ad67282b0a285bcd2de77f80e
-size 791

 version https://git-lfs.github.com/spec/v1
+oid sha256:ed0dd15fc572d2cc2ec19b317245622b4256a8737cc9ba114529e925d3bf42f2
+size 793

Llama-2-7b-hf_chunk6.mlmodelc/metadata.json CHANGED Viewed

@@ -7,9 +7,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_x",
         "type" : "MultiArray"
       },
@@ -17,9 +17,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_0",
         "type" : "MultiArray"
       },
@@ -27,9 +27,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_1",
         "type" : "MultiArray"
       },
@@ -37,9 +37,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_k_cache_2",
         "type" : "MultiArray"
       },
@@ -47,9 +47,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_0",
         "type" : "MultiArray"
       },
@@ -57,9 +57,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_1",
         "type" : "MultiArray"
       },
@@ -67,9 +67,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 64]",
         "name" : "new_v_cache_2",
         "type" : "MultiArray"
       }
@@ -79,17 +79,18 @@
     ],
     "specificationVersion" : 7,
     "mlProgramOperationTypeHistogram" : {
-      "Concat" : 18,
-      "Ios16.rsqrt" : 6,
-      "Ios16.mul" : 63,
-      "SliceByIndex" : 12,
       "Ios16.constexprLutToDense" : 21,
       "Ios16.conv" : 21,
-      "Ios16.add" : 21,
-      "Ios16.reduceMean" : 6,
-      "Ios16.matmul" : 6,
-      "Ios16.softmax" : 3,
-      "Ios16.reshape" : 12,
       "Ios16.silu" : 3
     },
     "computePrecision" : "Mixed (Float16, Int32)",
@@ -108,16 +109,16 @@
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
       "com.github.apple.coremltools.source" : "torch==2.1.0",
-      "com.github.apple.coremltools.version" : "7.2"
     },
     "inputSchema" : [
       {
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
-        "shape" : "[1, 4096, 1, 64]",
         "name" : "x",
         "type" : "MultiArray"
       },
@@ -145,9 +146,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 1 × 64 × 512)",
         "shortDescription" : "",
-        "shape" : "[1, 1, 64, 512]",
         "name" : "mask",
         "type" : "MultiArray"
       },
@@ -155,9 +156,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_0",
         "type" : "MultiArray"
       },
@@ -165,9 +166,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_0",
         "type" : "MultiArray"
       },
@@ -175,9 +176,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_1",
         "type" : "MultiArray"
       },
@@ -185,9 +186,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_1",
         "type" : "MultiArray"
       },
@@ -195,9 +196,9 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "k_cache_2",
         "type" : "MultiArray"
       },
@@ -205,14 +206,14 @@
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
-        "formattedType" : "MultiArray (Float16 1 × 32 × 128 × 448)?",
         "shortDescription" : "",
-        "shape" : "[1, 32, 128, 448]",
         "name" : "v_cache_2",
         "type" : "MultiArray"
       }
     ],
-    "generatedClassName" : "Llama_2_7b_hf_2024_05_25_14_03_55_chunk6",
     "method" : "predict"
   }
 ]

         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 8 × 8)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 8, 8]",
         "name" : "new_x",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 64 × 1 × 4096)",
         "shortDescription" : "",
+        "shape" : "[1, 64, 1, 4096]",
         "name" : "new_k_cache_2",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 64]",
         "name" : "new_v_cache_2",
         "type" : "MultiArray"
       }
     ],
     "specificationVersion" : 7,
     "mlProgramOperationTypeHistogram" : {
+      "Concat" : 21,
+      "Ios16.mul" : 150,
+      "SliceByIndex" : 300,
       "Ios16.constexprLutToDense" : 21,
+      "Transpose" : 3,
+      "Ios16.einsum" : 192,
       "Ios16.conv" : 21,
+      "Ios16.add" : 108,
+      "Ios16.realDiv" : 6,
+      "Ios16.softmax" : 96,
+      "Ios16.reduceL2Norm" : 6,
+      "Ios16.reshape" : 21,
       "Ios16.silu" : 3
     },
     "computePrecision" : "Mixed (Float16, Int32)",
     "userDefinedMetadata" : {
       "com.github.apple.coremltools.source_dialect" : "TorchScript",
       "com.github.apple.coremltools.source" : "torch==2.1.0",
+      "com.github.apple.coremltools.version" : "8.0b1"
     },
     "inputSchema" : [
       {
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 8 × 8)",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 8, 8]",
         "name" : "x",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "0",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 512 × 1 × 64)",
         "shortDescription" : "",
+        "shape" : "[1, 512, 1, 64]",
         "name" : "mask",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_0",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_1",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 448 × 1 × 4096)?",
         "shortDescription" : "",
+        "shape" : "[1, 448, 1, 4096]",
         "name" : "k_cache_2",
         "type" : "MultiArray"
       },
         "hasShapeFlexibility" : "0",
         "isOptional" : "1",
         "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 4096 × 1 × 448)?",
         "shortDescription" : "",
+        "shape" : "[1, 4096, 1, 448]",
         "name" : "v_cache_2",
         "type" : "MultiArray"
       }
     ],
+    "generatedClassName" : "Llama_2_7b_hf_2024_08_09_09_54_41_chunk6",
     "method" : "predict"
   }
 ]

Llama-2-7b-hf_chunk6.mlmodelc/model.mil CHANGED Viewed

The diff for this file is too large to render. See raw diff

Llama-2-7b-hf_chunk6.mlmodelc/weights/weight.bin CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:646d17c5d6d62e055abb88615254cb2d8205cd46a7b98faa734136f30c8ca26a
-size 303872704

 version https://git-lfs.github.com/spec/v1
+oid sha256:52ba7c726d6d195f52b6a623161d59bb42debc6a3e217eaf3f34bcd9be1b3834
+size 303873856