Multifunction chunks: SDPA decode + N=512 prefill, 50% weight dedup

mf/chunk1.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86b5fd7f5deb4db134d34e36554828f8a41fb9d5d8a03d2e5998e279f897f77f
+size 243

mf/chunk1.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b145aa3b71ec1af8e83efd69905de1cb606e9d75a51cf0e6ae5481bb0e516bf
+size 1360

mf/chunk1.mlmodelc/metadata.json

@@ -0,0 +1,760 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Mixed (Float16, Palettized (10 bits), Palettized (11 bits), Palettized (12 bits), Palettized (13 bits), Palettized (7 bits), Palettized (8 bits), UInt4)",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1536)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1536]",
+        "name" : "hidden_states_out",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 7 × 1 × 512 × 512)",
+        "shortDescription" : "",
+        "shape" : "[7, 1, 512, 512]",
+        "name" : "K_sliding_out",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 7 × 1 × 512 × 512)",
+        "shortDescription" : "",
+        "shape" : "[7, 1, 512, 512]",
+        "name" : "V_sliding_out",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 2048, 512]",
+        "name" : "K_full_out",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 2048, 512]",
+        "name" : "V_full_out",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 8960)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 8960]",
+        "name" : "per_layer_combined_out",
+        "type" : "MultiArray"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 9,
+    "functions" : [
+      {
+        "inputSchema" : [
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1536)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1536]",
+            "name" : "hidden_states",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 2048)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 2048]",
+            "name" : "causal_mask_full",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 512]",
+            "name" : "causal_mask_sliding",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 1)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 2048, 1]",
+            "name" : "update_mask",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 8960)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 8960]",
+            "name" : "per_layer_raw",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 256]",
+            "name" : "cos_s",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 256]",
+            "name" : "sin_s",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 512]",
+            "name" : "cos_f",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 512]",
+            "name" : "sin_f",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 7 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[7, 1, 512, 512]",
+            "name" : "K_sliding_in",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 7 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[7, 1, 512, 512]",
+            "name" : "V_sliding_in",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 2048, 512]",
+            "name" : "K_full_in",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 2048, 512]",
+            "name" : "V_full_in",
+            "type" : "MultiArray"
+          }
+        ],
+        "computePrecision" : "Mixed (Float16, Float32, Int32)",
+        "storagePrecision" : "Mixed (Float16, Palettized (10 bits), Palettized (11 bits), Palettized (12 bits), Palettized (13 bits), Palettized (7 bits), Palettized (8 bits), UInt4)",
+        "stateSchema" : [
+
+        ],
+        "outputSchema" : [
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1536)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1536]",
+            "name" : "hidden_states_out",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 7 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[7, 1, 512, 512]",
+            "name" : "K_sliding_out",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 7 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[7, 1, 512, 512]",
+            "name" : "V_sliding_out",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 2048, 512]",
+            "name" : "K_full_out",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 2048, 512]",
+            "name" : "V_full_out",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 8960)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 8960]",
+            "name" : "per_layer_combined_out",
+            "type" : "MultiArray"
+          }
+        ],
+        "name" : "decode",
+        "mlProgramOperationTypeHistogram" : {
+          "Ios18.expandDims" : 49,
+          "Ios18.mul" : 268,
+          "Ios18.rsqrt" : 8,
+          "Identity" : 2,
+          "Ios16.reduceMean" : 8,
+          "Ios18.scaledDotProductAttention" : 8,
+          "Split" : 107,
+          "Tile" : 18,
+          "Ios18.add" : 51,
+          "Ios18.layerNorm" : 91,
+          "Ios18.reshape" : 96,
+          "Pad" : 14,
+          "Ios18.constexprLutToDense" : 73,
+          "Ios18.conv" : 73,
+          "Ios18.gelu" : 16,
+          "Ios18.concat" : 122,
+          "Ios18.sub" : 1,
+          "Ios18.transpose" : 146,
+          "Ios18.pow" : 8,
+          "Stack" : 2,
+          "Ios18.sliceByIndex" : 87,
+          "Ios18.squeeze" : 31
+        }
+      },
+      {
+        "inputSchema" : [
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 512 × 1536)",
+            "shortDescription" : "",
+            "shape" : "[1, 512, 1536]",
+            "name" : "hidden_states",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "causal_mask",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 512 × 8960)",
+            "shortDescription" : "",
+            "shape" : "[1, 512, 8960]",
+            "name" : "per_layer_raw",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "cos_s",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "sin_s",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "cos_f",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "sin_f",
+            "type" : "MultiArray"
+          }
+        ],
+        "computePrecision" : "Mixed (Float16, Float32, Int32)",
+        "storagePrecision" : "Mixed (Float16, Palettized (10 bits), Palettized (11 bits), Palettized (12 bits), Palettized (13 bits), Palettized (7 bits), Palettized (8 bits), UInt4)",
+        "stateSchema" : [
+
+        ],
+        "outputSchema" : [
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 512 × 1536)",
+            "shortDescription" : "",
+            "shape" : "[1, 512, 1536]",
+            "name" : "hidden_states_out",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 512 × 8960)",
+            "shortDescription" : "",
+            "shape" : "[1, 512, 8960]",
+            "name" : "per_layer_combined_out",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "K0",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "V0",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "K1",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "V1",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "K2",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "V2",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "K3",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "V3",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "K4",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "V4",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "K5",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "V5",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "K6",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "V6",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "K7",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "V7",
+            "type" : "MultiArray"
+          }
+        ],
+        "name" : "prefill",
+        "mlProgramOperationTypeHistogram" : {
+          "Ios18.constexprLutToDense" : 73,
+          "Ios18.conv" : 73,
+          "Ios18.expandDims" : 33,
+          "Ios18.scaledDotProductAttention" : 8,
+          "Ios18.gelu" : 16,
+          "Ios18.concat" : 108,
+          "Ios18.add" : 49,
+          "Ios18.rsqrt" : 8,
+          "Ios16.reduceMean" : 8,
+          "Ios18.sliceByIndex" : 43,
+          "Ios18.layerNorm" : 91,
+          "Ios18.transpose" : 162,
+          "Ios18.pow" : 8,
+          "Ios18.squeeze" : 17,
+          "Ios18.reshape" : 96,
+          "Split" : 107,
+          "Tile" : 16,
+          "Ios18.mul" : 264
+        }
+      }
+    ],
+    "mlProgramOperationTypeHistogram" : {
+      "Ios18.expandDims" : 49,
+      "Ios18.mul" : 268,
+      "Ios18.rsqrt" : 8,
+      "Identity" : 2,
+      "Ios16.reduceMean" : 8,
+      "Ios18.scaledDotProductAttention" : 8,
+      "Split" : 107,
+      "Tile" : 18,
+      "Ios18.add" : 51,
+      "Ios18.layerNorm" : 91,
+      "Ios18.reshape" : 96,
+      "Pad" : 14,
+      "Ios18.constexprLutToDense" : 73,
+      "Ios18.conv" : 73,
+      "Ios18.gelu" : 16,
+      "Ios18.concat" : 122,
+      "Ios18.sub" : 1,
+      "Ios18.transpose" : 146,
+      "Ios18.pow" : 8,
+      "Stack" : 2,
+      "Ios18.sliceByIndex" : 87,
+      "Ios18.squeeze" : 31
+    },
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "availability" : {
+      "macOS" : "15.0",
+      "tvOS" : "18.0",
+      "visionOS" : "2.0",
+      "watchOS" : "11.0",
+      "iOS" : "18.0",
+      "macCatalyst" : "18.0"
+    },
+    "computePrecision" : "Mixed (Float16, Float32, Int32)",
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1536)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1536]",
+        "name" : "hidden_states",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 2048)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 2048]",
+        "name" : "causal_mask_full",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 512]",
+        "name" : "causal_mask_sliding",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 2048, 1]",
+        "name" : "update_mask",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 8960)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 8960]",
+        "name" : "per_layer_raw",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 256]",
+        "name" : "cos_s",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 256]",
+        "name" : "sin_s",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 512]",
+        "name" : "cos_f",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 512]",
+        "name" : "sin_f",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 7 × 1 × 512 × 512)",
+        "shortDescription" : "",
+        "shape" : "[7, 1, 512, 512]",
+        "name" : "K_sliding_in",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 7 × 1 × 512 × 512)",
+        "shortDescription" : "",
+        "shape" : "[7, 1, 512, 512]",
+        "name" : "V_sliding_in",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 2048, 512]",
+        "name" : "K_full_in",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 2048, 512]",
+        "name" : "V_full_in",
+        "type" : "MultiArray"
+      }
+    ],
+    "defaultFunctionName" : "decode",
+    "generatedClassName" : "chunk1",
+    "userDefinedMetadata" : {
+
+    },
+    "method" : "predict"
+  }
+]

mf/chunk1.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c9eaf1dccf8c3455368e013fc13807ff69c87f68272de66d7fef4b18970419e9
+size 155436864

mf/chunk2.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9fc331984cad6567889cc1fe875468eea83c07ef2f60884c6b0580618a13f0c6
+size 243

mf/chunk2.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eefa5ca2620252fbf47f18a2444f4b5ab57dbb7ef2515d3264f0b200c63cf8ea
+size 1365

mf/chunk2.mlmodelc/metadata.json

@@ -0,0 +1,788 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Mixed (Float16, Palettized (10 bits), Palettized (11 bits), Palettized (12 bits), Palettized (7 bits), Palettized (8 bits), UInt4)",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1536)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1536]",
+        "name" : "hidden_states_out",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 5 × 1 × 512 × 512)",
+        "shortDescription" : "",
+        "shape" : "[5, 1, 512, 512]",
+        "name" : "K_sliding_out",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 5 × 1 × 512 × 512)",
+        "shortDescription" : "",
+        "shape" : "[5, 1, 512, 512]",
+        "name" : "V_sliding_out",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 2 × 1 × 2048 × 512)",
+        "shortDescription" : "",
+        "shape" : "[2, 1, 2048, 512]",
+        "name" : "K_full_out",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 2 × 1 × 2048 × 512)",
+        "shortDescription" : "",
+        "shape" : "[2, 1, 2048, 512]",
+        "name" : "V_full_out",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 512, 256]",
+        "name" : "kv13_k",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 512, 256]",
+        "name" : "kv13_v",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 2048, 512]",
+        "name" : "kv14_k",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 2048, 512]",
+        "name" : "kv14_v",
+        "type" : "MultiArray"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 9,
+    "functions" : [
+      {
+        "inputSchema" : [
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1536)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1536]",
+            "name" : "hidden_states",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 2048)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 2048]",
+            "name" : "causal_mask_full",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 512]",
+            "name" : "causal_mask_sliding",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 1)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 2048, 1]",
+            "name" : "update_mask",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 8960)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 8960]",
+            "name" : "per_layer_combined",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 256]",
+            "name" : "cos_s",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 256]",
+            "name" : "sin_s",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 512]",
+            "name" : "cos_f",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 512]",
+            "name" : "sin_f",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 5 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[5, 1, 512, 512]",
+            "name" : "K_sliding_in",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 5 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[5, 1, 512, 512]",
+            "name" : "V_sliding_in",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 2 × 1 × 2048 × 512)",
+            "shortDescription" : "",
+            "shape" : "[2, 1, 2048, 512]",
+            "name" : "K_full_in",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 2 × 1 × 2048 × 512)",
+            "shortDescription" : "",
+            "shape" : "[2, 1, 2048, 512]",
+            "name" : "V_full_in",
+            "type" : "MultiArray"
+          }
+        ],
+        "computePrecision" : "Mixed (Float16, Float32, Int32)",
+        "storagePrecision" : "Mixed (Float16, Palettized (10 bits), Palettized (11 bits), Palettized (12 bits), Palettized (7 bits), Palettized (8 bits), UInt4)",
+        "stateSchema" : [
+
+        ],
+        "outputSchema" : [
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1536)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1536]",
+            "name" : "hidden_states_out",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 5 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[5, 1, 512, 512]",
+            "name" : "K_sliding_out",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 5 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[5, 1, 512, 512]",
+            "name" : "V_sliding_out",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 2 × 1 × 2048 × 512)",
+            "shortDescription" : "",
+            "shape" : "[2, 1, 2048, 512]",
+            "name" : "K_full_out",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 2 × 1 × 2048 × 512)",
+            "shortDescription" : "",
+            "shape" : "[2, 1, 2048, 512]",
+            "name" : "V_full_out",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "kv13_k",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "kv13_v",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 2048, 512]",
+            "name" : "kv14_k",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 2048, 512]",
+            "name" : "kv14_v",
+            "type" : "MultiArray"
+          }
+        ],
+        "name" : "decode",
+        "mlProgramOperationTypeHistogram" : {
+          "Ios18.expandDims" : 42,
+          "Ios18.mul" : 176,
+          "Ios18.rsqrt" : 7,
+          "Ios16.reduceMean" : 7,
+          "Ios18.scaledDotProductAttention" : 7,
+          "Split" : 63,
+          "Tile" : 18,
+          "Ios18.add" : 46,
+          "Ios18.layerNorm" : 49,
+          "Ios18.reshape" : 84,
+          "Pad" : 10,
+          "Ios18.constexprLutToDense" : 63,
+          "Ios18.conv" : 63,
+          "Ios18.gelu" : 14,
+          "Ios18.concat" : 73,
+          "Ios18.sub" : 1,
+          "Ios18.pow" : 7,
+          "Ios18.transpose" : 126,
+          "Stack" : 4,
+          "Ios18.sliceByIndex" : 41,
+          "Ios18.squeeze" : 28
+        }
+      },
+      {
+        "inputSchema" : [
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 512 × 1536)",
+            "shortDescription" : "",
+            "shape" : "[1, 512, 1536]",
+            "name" : "hidden_states",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "causal_mask",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 512 × 8960)",
+            "shortDescription" : "",
+            "shape" : "[1, 512, 8960]",
+            "name" : "per_layer_combined",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "cos_s",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "sin_s",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "cos_f",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "sin_f",
+            "type" : "MultiArray"
+          }
+        ],
+        "computePrecision" : "Mixed (Float16, Float32, Int32)",
+        "storagePrecision" : "Mixed (Float16, Palettized (10 bits), Palettized (11 bits), Palettized (12 bits), Palettized (7 bits), Palettized (8 bits), UInt4)",
+        "stateSchema" : [
+
+        ],
+        "outputSchema" : [
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 512 × 1536)",
+            "shortDescription" : "",
+            "shape" : "[1, 512, 1536]",
+            "name" : "hidden_states_out",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "K0",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "V0",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "K1",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "V1",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "K2",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "V2",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "K3",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "V3",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "K4",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "V4",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "kv13_k",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "kv13_v",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "kv14_k",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "kv14_v",
+            "type" : "MultiArray"
+          }
+        ],
+        "name" : "prefill",
+        "mlProgramOperationTypeHistogram" : {
+          "Ios18.constexprLutToDense" : 63,
+          "Ios18.conv" : 63,
+          "Ios18.expandDims" : 28,
+          "Ios18.scaledDotProductAttention" : 7,
+          "Ios18.gelu" : 14,
+          "Ios18.concat" : 63,
+          "Ios18.add" : 42,
+          "Ios18.rsqrt" : 7,
+          "Ios16.reduceMean" : 7,
+          "Ios18.pow" : 7,
+          "Ios18.transpose" : 140,
+          "Ios18.layerNorm" : 49,
+          "Tile" : 14,
+          "Ios18.squeeze" : 14,
+          "Ios18.reshape" : 84,
+          "Split" : 63,
+          "Ios18.sliceByIndex" : 7,
+          "Ios18.mul" : 168
+        }
+      }
+    ],
+    "mlProgramOperationTypeHistogram" : {
+      "Ios18.expandDims" : 42,
+      "Ios18.mul" : 176,
+      "Ios18.rsqrt" : 7,
+      "Ios16.reduceMean" : 7,
+      "Ios18.scaledDotProductAttention" : 7,
+      "Split" : 63,
+      "Tile" : 18,
+      "Ios18.add" : 46,
+      "Ios18.layerNorm" : 49,
+      "Ios18.reshape" : 84,
+      "Pad" : 10,
+      "Ios18.constexprLutToDense" : 63,
+      "Ios18.conv" : 63,
+      "Ios18.gelu" : 14,
+      "Ios18.concat" : 73,
+      "Ios18.sub" : 1,
+      "Ios18.pow" : 7,
+      "Ios18.transpose" : 126,
+      "Stack" : 4,
+      "Ios18.sliceByIndex" : 41,
+      "Ios18.squeeze" : 28
+    },
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "availability" : {
+      "macOS" : "15.0",
+      "tvOS" : "18.0",
+      "visionOS" : "2.0",
+      "watchOS" : "11.0",
+      "iOS" : "18.0",
+      "macCatalyst" : "18.0"
+    },
+    "computePrecision" : "Mixed (Float16, Float32, Int32)",
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1536)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1536]",
+        "name" : "hidden_states",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 2048)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 2048]",
+        "name" : "causal_mask_full",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 512]",
+        "name" : "causal_mask_sliding",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 2048, 1]",
+        "name" : "update_mask",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 8960)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 8960]",
+        "name" : "per_layer_combined",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 256]",
+        "name" : "cos_s",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 256]",
+        "name" : "sin_s",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 512]",
+        "name" : "cos_f",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 512]",
+        "name" : "sin_f",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 5 × 1 × 512 × 512)",
+        "shortDescription" : "",
+        "shape" : "[5, 1, 512, 512]",
+        "name" : "K_sliding_in",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 5 × 1 × 512 × 512)",
+        "shortDescription" : "",
+        "shape" : "[5, 1, 512, 512]",
+        "name" : "V_sliding_in",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 2 × 1 × 2048 × 512)",
+        "shortDescription" : "",
+        "shape" : "[2, 1, 2048, 512]",
+        "name" : "K_full_in",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 2 × 1 × 2048 × 512)",
+        "shortDescription" : "",
+        "shape" : "[2, 1, 2048, 512]",
+        "name" : "V_full_in",
+        "type" : "MultiArray"
+      }
+    ],
+    "defaultFunctionName" : "decode",
+    "generatedClassName" : "chunk2",
+    "userDefinedMetadata" : {
+
+    },
+    "method" : "predict"
+  }
+]

mf/chunk2.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c25d52a035e531bf37a39688d52b49456541ba9d0c32f45850a6fbe7acfcab1b
+size 133964544

mf/chunk3.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e411899900c12d53665b02f91c599382019f4b403a2cfb4ff608a63381152823
+size 243

mf/chunk3.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:844e182dafa176af20e67a4d1f86be9c57582878a967b469bf5eaa6dc50685d5
+size 897

mf/chunk3.mlmodelc/metadata.json

@@ -0,0 +1,515 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Mixed (Float16, Palettized (10 bits), Palettized (11 bits), Palettized (13 bits), Palettized (7 bits), UInt4)",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1536)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1536]",
+        "name" : "hidden_states_out",
+        "type" : "MultiArray"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 9,
+    "functions" : [
+      {
+        "inputSchema" : [
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1536)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1536]",
+            "name" : "hidden_states",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 2048)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 2048]",
+            "name" : "causal_mask_full",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 512]",
+            "name" : "causal_mask_sliding",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 1)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 2048, 1]",
+            "name" : "update_mask",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 8960)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 8960]",
+            "name" : "per_layer_combined",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 256]",
+            "name" : "cos_s",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 256]",
+            "name" : "sin_s",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 512]",
+            "name" : "cos_f",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 512]",
+            "name" : "sin_f",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "kv13_k",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "kv13_v",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 2048, 512]",
+            "name" : "kv14_k",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 2048, 512]",
+            "name" : "kv14_v",
+            "type" : "MultiArray"
+          }
+        ],
+        "computePrecision" : "Mixed (Float16, Int32)",
+        "storagePrecision" : "Mixed (Float16, Palettized (10 bits), Palettized (11 bits), Palettized (13 bits), Palettized (7 bits), UInt4)",
+        "stateSchema" : [
+
+        ],
+        "outputSchema" : [
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1536)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1536]",
+            "name" : "hidden_states_out",
+            "type" : "MultiArray"
+          }
+        ],
+        "name" : "decode",
+        "mlProgramOperationTypeHistogram" : {
+          "Ios18.constexprLutToDense" : 70,
+          "Ios18.scaledDotProductAttention" : 10,
+          "Ios18.expandDims" : 40,
+          "Ios18.conv" : 70,
+          "Ios18.gelu" : 20,
+          "Ios18.concat" : 70,
+          "Ios18.add" : 40,
+          "Ios18.sliceByIndex" : 10,
+          "Tile" : 4,
+          "Ios18.transpose" : 112,
+          "Ios18.layerNorm" : 60,
+          "Ios18.squeeze" : 20,
+          "Ios18.reshape" : 48,
+          "Split" : 70,
+          "Identity" : 1,
+          "Ios18.mul" : 180
+        }
+      },
+      {
+        "inputSchema" : [
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 512 × 1536)",
+            "shortDescription" : "",
+            "shape" : "[1, 512, 1536]",
+            "name" : "hidden_states",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "causal_mask",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 512 × 8960)",
+            "shortDescription" : "",
+            "shape" : "[1, 512, 8960]",
+            "name" : "per_layer_combined",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "cos_s",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "sin_s",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "cos_f",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "sin_f",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "kv13_k",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "kv13_v",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "kv14_k",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "kv14_v",
+            "type" : "MultiArray"
+          }
+        ],
+        "computePrecision" : "Mixed (Float16, Int32)",
+        "storagePrecision" : "Mixed (Float16, Palettized (10 bits), Palettized (11 bits), Palettized (13 bits), Palettized (7 bits), UInt4)",
+        "stateSchema" : [
+
+        ],
+        "outputSchema" : [
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 512 × 1536)",
+            "shortDescription" : "",
+            "shape" : "[1, 512, 1536]",
+            "name" : "hidden_states_out",
+            "type" : "MultiArray"
+          }
+        ],
+        "name" : "prefill",
+        "mlProgramOperationTypeHistogram" : {
+          "Ios18.constexprLutToDense" : 70,
+          "Ios18.scaledDotProductAttention" : 10,
+          "Ios18.expandDims" : 40,
+          "Ios18.conv" : 70,
+          "Ios18.gelu" : 20,
+          "Ios18.concat" : 70,
+          "Ios18.add" : 40,
+          "Ios18.sliceByIndex" : 10,
+          "Tile" : 4,
+          "Ios18.transpose" : 122,
+          "Ios18.layerNorm" : 60,
+          "Ios18.squeeze" : 20,
+          "Ios18.reshape" : 48,
+          "Split" : 70,
+          "Ios18.mul" : 180
+        }
+      }
+    ],
+    "mlProgramOperationTypeHistogram" : {
+      "Ios18.constexprLutToDense" : 70,
+      "Ios18.scaledDotProductAttention" : 10,
+      "Ios18.expandDims" : 40,
+      "Ios18.conv" : 70,
+      "Ios18.gelu" : 20,
+      "Ios18.concat" : 70,
+      "Ios18.add" : 40,
+      "Ios18.sliceByIndex" : 10,
+      "Tile" : 4,
+      "Ios18.transpose" : 112,
+      "Ios18.layerNorm" : 60,
+      "Ios18.squeeze" : 20,
+      "Ios18.reshape" : 48,
+      "Split" : 70,
+      "Identity" : 1,
+      "Ios18.mul" : 180
+    },
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "availability" : {
+      "macOS" : "15.0",
+      "tvOS" : "18.0",
+      "visionOS" : "2.0",
+      "watchOS" : "11.0",
+      "iOS" : "18.0",
+      "macCatalyst" : "18.0"
+    },
+    "computePrecision" : "Mixed (Float16, Int32)",
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1536)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1536]",
+        "name" : "hidden_states",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 2048)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 2048]",
+        "name" : "causal_mask_full",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 512]",
+        "name" : "causal_mask_sliding",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 2048, 1]",
+        "name" : "update_mask",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 8960)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 8960]",
+        "name" : "per_layer_combined",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 256]",
+        "name" : "cos_s",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 256]",
+        "name" : "sin_s",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 512]",
+        "name" : "cos_f",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 512]",
+        "name" : "sin_f",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 512, 256]",
+        "name" : "kv13_k",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 512, 256]",
+        "name" : "kv13_v",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 2048, 512]",
+        "name" : "kv14_k",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 2048, 512]",
+        "name" : "kv14_v",
+        "type" : "MultiArray"
+      }
+    ],
+    "defaultFunctionName" : "decode",
+    "generatedClassName" : "chunk3",
+    "userDefinedMetadata" : {
+
+    },
+    "method" : "predict"
+  }
+]

mf/chunk3.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:560ba9d7ba21739817d78aa46e00abb5b14b5dbb12ebdd6f2736515b0cbf8672
+size 325282880

mf/chunk4.mlmodelc/analytics/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9524736efd306828984931fba24e4180bbdd5cb3203ebb0e3e5940a5afb49aa8
+size 243

mf/chunk4.mlmodelc/coremldata.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c17444352416692476cf4d3cdae49b738e323ea785d303df273d2392338c5ac5
+size 960

mf/chunk4.mlmodelc/metadata.json

@@ -0,0 +1,565 @@
+[
+  {
+    "metadataOutputVersion" : "3.0",
+    "storagePrecision" : "Mixed (Float16, Palettized (10 bits), Palettized (11 bits), Palettized (13 bits), Palettized (17 bits), Palettized (7 bits), UInt4)",
+    "outputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Int32",
+        "formattedType" : "MultiArray (Int32 1)",
+        "shortDescription" : "",
+        "shape" : "[1]",
+        "name" : "token_id",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1)",
+        "shortDescription" : "",
+        "shape" : "[1]",
+        "name" : "token_logit",
+        "type" : "MultiArray"
+      }
+    ],
+    "modelParameters" : [
+
+    ],
+    "specificationVersion" : 9,
+    "functions" : [
+      {
+        "inputSchema" : [
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1536)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1536]",
+            "name" : "hidden_states",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 2048)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 2048]",
+            "name" : "causal_mask_full",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 512]",
+            "name" : "causal_mask_sliding",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 1)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 2048, 1]",
+            "name" : "update_mask",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 8960)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 8960]",
+            "name" : "per_layer_combined",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 256]",
+            "name" : "cos_s",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 256]",
+            "name" : "sin_s",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 512]",
+            "name" : "cos_f",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 1, 512]",
+            "name" : "sin_f",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "kv13_k",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "kv13_v",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 2048, 512]",
+            "name" : "kv14_k",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 2048, 512]",
+            "name" : "kv14_v",
+            "type" : "MultiArray"
+          }
+        ],
+        "computePrecision" : "Mixed (Float16, Int32)",
+        "storagePrecision" : "Mixed (Float16, Palettized (10 bits), Palettized (11 bits), Palettized (13 bits), Palettized (17 bits), Palettized (7 bits), UInt4)",
+        "stateSchema" : [
+
+        ],
+        "outputSchema" : [
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Int32",
+            "formattedType" : "MultiArray (Int32 1)",
+            "shortDescription" : "",
+            "shape" : "[1]",
+            "name" : "token_id",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1)",
+            "shortDescription" : "",
+            "shape" : "[1]",
+            "name" : "token_logit",
+            "type" : "MultiArray"
+          }
+        ],
+        "name" : "decode",
+        "mlProgramOperationTypeHistogram" : {
+          "Ios18.constexprLutToDense" : 71,
+          "Identity" : 1,
+          "Ios18.scaledDotProductAttention" : 10,
+          "Ios18.expandDims" : 41,
+          "Ios18.conv" : 71,
+          "Ios18.reduceArgmax" : 1,
+          "Ios18.gelu" : 20,
+          "Ios18.concat" : 71,
+          "Ios18.add" : 40,
+          "Ios18.tanh" : 1,
+          "Ios18.sliceByIndex" : 10,
+          "Tile" : 4,
+          "Ios18.transpose" : 114,
+          "Ios18.layerNorm" : 61,
+          "Ios18.squeeze" : 22,
+          "Ios18.reshape" : 48,
+          "Split" : 71,
+          "Ios18.gatherAlongAxis" : 1,
+          "Ios18.mul" : 184
+        }
+      },
+      {
+        "inputSchema" : [
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 512 × 1536)",
+            "shortDescription" : "",
+            "shape" : "[1, 512, 1536]",
+            "name" : "hidden_states",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "causal_mask",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 512 × 8960)",
+            "shortDescription" : "",
+            "shape" : "[1, 512, 8960]",
+            "name" : "per_layer_combined",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "cos_s",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "sin_s",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "cos_f",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "sin_f",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "kv13_k",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 256]",
+            "name" : "kv13_v",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "kv14_k",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 512)",
+            "shortDescription" : "",
+            "shape" : "[1, 1, 512, 512]",
+            "name" : "kv14_v",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1 × 512 × 1)",
+            "shortDescription" : "",
+            "shape" : "[1, 512, 1]",
+            "name" : "last_position_mask",
+            "type" : "MultiArray"
+          }
+        ],
+        "computePrecision" : "Mixed (Float16, Int32)",
+        "storagePrecision" : "Mixed (Float16, Palettized (10 bits), Palettized (11 bits), Palettized (13 bits), Palettized (17 bits), Palettized (7 bits), UInt4)",
+        "stateSchema" : [
+
+        ],
+        "outputSchema" : [
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Int32",
+            "formattedType" : "MultiArray (Int32 1)",
+            "shortDescription" : "",
+            "shape" : "[1]",
+            "name" : "token_id",
+            "type" : "MultiArray"
+          },
+          {
+            "hasShapeFlexibility" : "0",
+            "isOptional" : "0",
+            "dataType" : "Float16",
+            "formattedType" : "MultiArray (Float16 1)",
+            "shortDescription" : "",
+            "shape" : "[1]",
+            "name" : "token_logit",
+            "type" : "MultiArray"
+          }
+        ],
+        "name" : "prefill",
+        "mlProgramOperationTypeHistogram" : {
+          "Ios18.constexprLutToDense" : 71,
+          "Ios18.gatherAlongAxis" : 1,
+          "Ios18.scaledDotProductAttention" : 10,
+          "Ios18.expandDims" : 41,
+          "Ios18.conv" : 71,
+          "Ios18.reduceArgmax" : 1,
+          "Ios18.gelu" : 20,
+          "Ios18.concat" : 71,
+          "Ios18.add" : 40,
+          "Ios18.tanh" : 1,
+          "Ios18.sliceByIndex" : 10,
+          "Tile" : 4,
+          "Ios18.transpose" : 124,
+          "Ios18.layerNorm" : 61,
+          "Ios18.squeeze" : 22,
+          "Ios18.reshape" : 48,
+          "Split" : 71,
+          "Ios16.reduceSum" : 1,
+          "Ios18.mul" : 185
+        }
+      }
+    ],
+    "mlProgramOperationTypeHistogram" : {
+      "Ios18.constexprLutToDense" : 71,
+      "Identity" : 1,
+      "Ios18.scaledDotProductAttention" : 10,
+      "Ios18.expandDims" : 41,
+      "Ios18.conv" : 71,
+      "Ios18.reduceArgmax" : 1,
+      "Ios18.gelu" : 20,
+      "Ios18.concat" : 71,
+      "Ios18.add" : 40,
+      "Ios18.tanh" : 1,
+      "Ios18.sliceByIndex" : 10,
+      "Tile" : 4,
+      "Ios18.transpose" : 114,
+      "Ios18.layerNorm" : 61,
+      "Ios18.squeeze" : 22,
+      "Ios18.reshape" : 48,
+      "Split" : 71,
+      "Ios18.gatherAlongAxis" : 1,
+      "Ios18.mul" : 184
+    },
+    "isUpdatable" : "0",
+    "stateSchema" : [
+
+    ],
+    "availability" : {
+      "macOS" : "15.0",
+      "tvOS" : "18.0",
+      "visionOS" : "2.0",
+      "watchOS" : "11.0",
+      "iOS" : "18.0",
+      "macCatalyst" : "18.0"
+    },
+    "computePrecision" : "Mixed (Float16, Int32)",
+    "modelType" : {
+      "name" : "MLModelType_mlProgram"
+    },
+    "inputSchema" : [
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1536)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1536]",
+        "name" : "hidden_states",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 2048)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 2048]",
+        "name" : "causal_mask_full",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 512]",
+        "name" : "causal_mask_sliding",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 1)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 2048, 1]",
+        "name" : "update_mask",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 8960)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 8960]",
+        "name" : "per_layer_combined",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 256]",
+        "name" : "cos_s",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 256]",
+        "name" : "sin_s",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 512]",
+        "name" : "cos_f",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 1 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 1, 512]",
+        "name" : "sin_f",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 512, 256]",
+        "name" : "kv13_k",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 512 × 256)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 512, 256]",
+        "name" : "kv13_v",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 2048, 512]",
+        "name" : "kv14_k",
+        "type" : "MultiArray"
+      },
+      {
+        "hasShapeFlexibility" : "0",
+        "isOptional" : "0",
+        "dataType" : "Float16",
+        "formattedType" : "MultiArray (Float16 1 × 1 × 2048 × 512)",
+        "shortDescription" : "",
+        "shape" : "[1, 1, 2048, 512]",
+        "name" : "kv14_v",
+        "type" : "MultiArray"
+      }
+    ],
+    "defaultFunctionName" : "decode",
+    "generatedClassName" : "chunk4",
+    "userDefinedMetadata" : {
+
+    },
+    "method" : "predict"
+  }
+]

mf/chunk4.mlmodelc/weights/weight.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3b34de9f847b88c27155b1c5acf361d74d7f42508062c7ed35a03666aa42315c
+size 526874880