Upload folder using huggingface_hub

config.json

@@ -42,6 +42,7 @@
   "max_length": 448,
   "max_source_positions": 1500,
   "max_target_positions": 448,
+  "median_filter_width": 7,
   "model_type": "whisper",
   "num_hidden_layers": 24,
   "num_mel_bins": 80,
@@ -139,7 +140,7 @@
     50360,
     50361
   ],
-  "transformers_version": "4.30.2",
+  "transformers_version": "4.33.0.dev0",
   "use_cache": true,
   "use_weighted_layer_sum": false,
   "vocab_size": 51864

generation_config.json

@@ -1,5 +1,78 @@
 {
-  "alignment_heads": [[11, 4], [14, 1], [14, 12], [14, 14], [15, 4], [16, 0], [16, 4], [16, 9], [17, 12], [17, 14], [18, 7], [18, 10], [18, 15], [20, 0], [20, 3], [20, 9], [20, 14], [21, 12]],
+  "alignment_heads": [
+    [
+      11,
+      4
+    ],
+    [
+      14,
+      1
+    ],
+    [
+      14,
+      12
+    ],
+    [
+      14,
+      14
+    ],
+    [
+      15,
+      4
+    ],
+    [
+      16,
+      0
+    ],
+    [
+      16,
+      4
+    ],
+    [
+      16,
+      9
+    ],
+    [
+      17,
+      12
+    ],
+    [
+      17,
+      14
+    ],
+    [
+      18,
+      7
+    ],
+    [
+      18,
+      10
+    ],
+    [
+      18,
+      15
+    ],
+    [
+      20,
+      0
+    ],
+    [
+      20,
+      3
+    ],
+    [
+      20,
+      9
+    ],
+    [
+      20,
+      14
+    ],
+    [
+      21,
+      12
+    ]
+  ],
   "begin_suppress_tokens": [
     220,
     50256
@@ -111,5 +184,5 @@
     50360,
     50361
   ],
-  "transformers_version": "4.30.2"
+  "transformers_version": "4.33.0.dev0"
 }

onnx/decoder_model.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6c768cbdc2f7da73f15400f34e4e20b439f043ab870bb3e80c7d2d9f40f242bd
-size 2040006900
+oid sha256:94d45552fa1245ee26fdd5189637d1bdda3b5db7e403fd7e4c173db90826e071
+size 1827703309

onnx/decoder_model_merged.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2a6bc26a22416cb5ca53f5db05a44fae39a56a38053bd172c60a77be35974151
-size 1926573
+oid sha256:9361cc5a61a932b6e478b57401f1fe55ef86181aa1ee6adcce121c36ce03bd6c
+size 1828724169

onnx/decoder_model_merged_quantized.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cd4c06f8bfbfb28f8a777a84618692ee05687ab598fef9dffb2fea11d3cd83c4
-size 515405408
+oid sha256:0b009dc2d71add45954b927ff01815d25d9d397bbb8b2ad72c6b9bd3279d7607
+size 462660582

onnx/decoder_model_quantized.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8fe4c89bbf88b377dd1d2862571d7611ce6888d8d1a055b61790050791d97deb
-size 514017234
+oid sha256:3e2f4fed175268b904c6daf9ffb59f095864aa9f3b0217dcdeaa88a5307759f8
+size 461078933

onnx/decoder_with_past_model.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0fd1ee78061a395fd1e623ec308a19d9b4803b450aa8e70b24fad833c17e3c76
-size 1838433559
+oid sha256:fce529ecc15e2abd10d50baafa927be37043764da0bb1832fb28a05b8e2ccac7
+size 1626142598

onnx/decoder_with_past_model_quantized.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:789bc74086187043db99dd1684f3d82ec61d15464dddc2c3daba21dec6e505d5
-size 463338655
+oid sha256:98c5f7b1868fb1cc0a4b3838652f41748784015e7a2675fbbc536ac1328c9d16
+size 410417655

onnx/encoder_model.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2846e5d8c8c847af016939d49add139b57f3a1f5a2739f8eb77f48d18f67e552
-size 1229258863
+oid sha256:f334bbbcdf56bb3e2a82cfb523673fdd4b7a6b5d9b7298e9d7fcefd03fe9ada5
+size 1229312445

onnx/encoder_model_quantized.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:952aacc7ab4ff86afccc732a42c4d6c8babb7651653aeb9b0ff22adcc42dcc90
-size 313397176
+oid sha256:c6fdaff8017b837b7b205ccd1d5f283008f4ee9677a29e503b3cf8e68efa9bb3
+size 313468028

quantize_config.json

@@ -0,0 +1,115 @@
+{
+    "per_channel": false,
+    "reduce_range": false,
+    "per_model_config": {
+        "decoder_model": {
+            "op_types": [
+                "Cast",
+                "Gather",
+                "Sqrt",
+                "Constant",
+                "Softmax",
+                "MatMul",
+                "Unsqueeze",
+                "Reshape",
+                "ConstantOfShape",
+                "Range",
+                "Add",
+                "Where",
+                "Equal",
+                "Transpose",
+                "Slice",
+                "Concat",
+                "Div",
+                "Less",
+                "ReduceMean",
+                "Squeeze",
+                "Shape",
+                "Erf",
+                "Sub",
+                "Mul",
+                "Expand",
+                "Pow"
+            ],
+            "weight_type": "QInt8"
+        },
+        "decoder_model_merged": {
+            "op_types": [
+                "Cast",
+                "Gather",
+                "Sqrt",
+                "Constant",
+                "Softmax",
+                "MatMul",
+                "Unsqueeze",
+                "Reshape",
+                "ConstantOfShape",
+                "Range",
+                "Add",
+                "Where",
+                "If",
+                "Equal",
+                "Transpose",
+                "Slice",
+                "Concat",
+                "Div",
+                "Less",
+                "ReduceMean",
+                "Squeeze",
+                "Shape",
+                "Erf",
+                "Sub",
+                "Mul",
+                "Expand",
+                "Pow"
+            ],
+            "weight_type": "QInt8"
+        },
+        "decoder_with_past_model": {
+            "op_types": [
+                "MatMul",
+                "Erf",
+                "Sub",
+                "Add",
+                "Unsqueeze",
+                "Softmax",
+                "Reshape",
+                "Mul",
+                "Transpose",
+                "Slice",
+                "Concat",
+                "Gather",
+                "Div",
+                "ReduceMean",
+                "Sqrt",
+                "Pow",
+                "Constant",
+                "Shape"
+            ],
+            "weight_type": "QInt8"
+        },
+        "encoder_model": {
+            "op_types": [
+                "MatMul",
+                "Erf",
+                "Sub",
+                "Add",
+                "Unsqueeze",
+                "Softmax",
+                "Reshape",
+                "Mul",
+                "Transpose",
+                "Div",
+                "Concat",
+                "Gather",
+                "Conv",
+                "ReduceMean",
+                "Sqrt",
+                "Pow",
+                "Constant",
+                "Shape"
+            ],
+            "weight_type": "QUInt8"
+        }
+    }
+}