updated convert script

Llama-3.2-3B-Instruct.mlmodelc/analytics/coremldata.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bce285869bcc6b73ec1c649c1d6807ecc19f56a6e5d76d14a9116a054a340a0a
+oid sha256:a3d4581f7d6f35d9f7a50bf30659554f8f43aa69048127270d2890edc6369677
 size 243

Llama-3.2-3B-Instruct.mlmodelc/coremldata.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9a517abd873ff25c8eeab2bfaeb2e852f904ef7143a9e0c38ed96095f63a1bef
+oid sha256:bd8ae4045a77075badabbfd6bf2e9d2c0e0ba26838955df07212af8178127ee2
 size 443

Llama-3.2-3B-Instruct.mlmodelc/metadata.json

@@ -111,7 +111,7 @@
         "shortDescription" : ""
       }
     ],
-    "generatedClassName" : "Llama_3_2_3B_Instruct_",
+    "generatedClassName" : "Llama_3_2_3B_Instruct",
     "method" : "predict"
   }
 ]

Llama-3.2-3B-Instruct.mlmodelc/model.mil

@@ -5763,8 +5763,8 @@ program(1.3)
             tensor<fp16, [1, ?, 1]> var_5171_cast_fp16 = rsqrt(epsilon = var_5171_epsilon_0, x = var_5170_cast_fp16)[name = string("op_5171_cast_fp16")];
             tensor<fp16, [1, ?, 3072]> hidden_states_843_cast_fp16 = mul(x = hidden_states_839_cast_fp16, y = var_5171_cast_fp16)[name = string("hidden_states_843_cast_fp16")];
             tensor<fp16, [3072]> model_model_norm_weight_to_fp16 = const()[name = string("model_model_norm_weight_to_fp16"), val = tensor<fp16, [3072]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(1807471168)))];
-            tensor<fp16, [1, ?, 3072]> hidden_states_cast_fp16 = mul(x = model_model_norm_weight_to_fp16, y = hidden_states_843_cast_fp16)[name = string("hidden_states_cast_fp16")];
+            tensor<fp16, [1, ?, 3072]> input_cast_fp16 = mul(x = model_model_norm_weight_to_fp16, y = hidden_states_843_cast_fp16)[name = string("input_cast_fp16")];
             tensor<fp16, [128256]> linear_196_bias_0_to_fp16 = const()[name = string("linear_196_bias_0_to_fp16"), val = tensor<fp16, [128256]>(BLOBFILE(path = string("@model_path/weights/weight.bin"), offset = uint64(1807477376)))];
-            tensor<fp16, [1, ?, 128256]> logits = linear(bias = linear_196_bias_0_to_fp16, weight = model_model_embed_tokens_weight_to_fp16_quantized, x = hidden_states_cast_fp16)[name = string("linear_196_cast_fp16")];
+            tensor<fp16, [1, ?, 128256]> logits = linear(bias = linear_196_bias_0_to_fp16, weight = model_model_embed_tokens_weight_to_fp16_quantized, x = input_cast_fp16)[name = string("linear_196_cast_fp16")];
         } -> (logits);
 }

{Llama-3.2-3B-Instruct-CoreML.mlpackage → Llama-3.2-3B-Instruct.mlpackage}/Data/com.apple.CoreML/model.mlmodel

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:197870b0474a71ac11c656b6283c6df04b48c7cd199e95db5471749bb89cc667
-size 944971
+oid sha256:a5d041bbd3b955595ff286ffa8737a25f519e0818b8152e0c7f0f19875de81d7
+size 944947

{Llama-3.2-3B-Instruct-CoreML.mlpackage → Llama-3.2-3B-Instruct.mlpackage}/Manifest.json

@@ -1,18 +1,18 @@
 {
     "fileFormatVersion": "1.0.0",
     "itemInfoEntries": {
-        "1D4E30CA-14B4-466E-84FD-C11B3FACC826": {
-            "author": "com.apple.CoreML",
-            "description": "CoreML Model Weights",
-            "name": "weights",
-            "path": "com.apple.CoreML/weights"
-        },
-        "6D00BF4D-7896-409E-9471-159CBF45A850": {
+        "48FD28CF-8BE6-4E1B-AAD6-F2ADD809E61C": {
             "author": "com.apple.CoreML",
             "description": "CoreML Model Specification",
             "name": "model.mlmodel",
             "path": "com.apple.CoreML/model.mlmodel"
+        },
+        "B399F7F1-41A3-45CC-8DCA-4999C63EA1F5": {
+            "author": "com.apple.CoreML",
+            "description": "CoreML Model Weights",
+            "name": "weights",
+            "path": "com.apple.CoreML/weights"
         }
     },
-    "rootModelIdentifier": "6D00BF4D-7896-409E-9471-159CBF45A850"
+    "rootModelIdentifier": "48FD28CF-8BE6-4E1B-AAD6-F2ADD809E61C"
 }