Upload folder using huggingface_hub

config.json

@@ -50,6 +50,7 @@
   "max_length": 448,
   "max_source_positions": 1500,
   "max_target_positions": 448,
+  "median_filter_width": 7,
   "model_type": "whisper",
   "num_hidden_layers": 12,
   "num_mel_bins": 80,
@@ -143,7 +144,7 @@
     50361,
     50362
   ],
-  "transformers_version": "4.30.2",
+  "transformers_version": "4.33.0.dev0",
   "use_cache": true,
   "use_weighted_layer_sum": false,
   "vocab_size": 51865

generation_config.json

@@ -1,5 +1,46 @@
 {
-  "alignment_heads": [[5, 3], [5, 9], [8, 0], [8, 4], [8, 7], [8, 8], [9, 0], [9, 7], [9, 9], [10, 5]],
+  "alignment_heads": [
+    [
+      5,
+      3
+    ],
+    [
+      5,
+      9
+    ],
+    [
+      8,
+      0
+    ],
+    [
+      8,
+      4
+    ],
+    [
+      8,
+      7
+    ],
+    [
+      8,
+      8
+    ],
+    [
+      9,
+      0
+    ],
+    [
+      9,
+      7
+    ],
+    [
+      9,
+      9
+    ],
+    [
+      10,
+      5
+    ]
+  ],
   "begin_suppress_tokens": [
     220,
     50257
@@ -218,5 +259,5 @@
     "transcribe": 50359,
     "translate": 50358
   },
-  "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 sha256:2cb0b4392373471d5bbfa056d48336b74907e1542583d069dd8fb8fa8868df46
-size 774159552
+oid sha256:a5612573a4cba5056778cb0cb9be51240085abda6f9c18d2de190fe026c71718
+size 614895553

onnx/decoder_model_merged.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c09ce6b24db9c8a55eab37aba77fb217052da64fa7f30d5f0411008644bc0e15
-size 774594491
+oid sha256:16c197b4c4702c51c767d1931a5673a10fa528bd1e79963891c97d69e0258584
+size 615405212

onnx/decoder_model_merged_quantized.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:df24179148e324c957c5740928e54c58f7894db9022949eb4ba4e5a5971aa4b9
-size 196432559
+oid sha256:fcfc6100dc7339e7507e10f8b274350be7c4f8d8b575f0293f94cc0e156d6d24
+size 156780950

onnx/decoder_model_quantized.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:48bed0449e061d9ceef7a1b6b54800e9c730478e769404309e34dca27e4a89f4
-size 195739168
+oid sha256:1514de20d1d3f177486c76260ee943849e483f4971bd7285d060050c2d67d485
+size 155991279

onnx/decoder_with_past_model.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4a3b5852abece73bd2fe405d80bbf44ade68acdaf80a7c236f1dc004e0b6d55f
-size 717422863
+oid sha256:29965c98dc6ba99eab95d23a884b37017235fd9890423f499d5e5d72baecb91d
+size 558165134

onnx/decoder_with_past_model_quantized.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f25f8c98570159bb7ed56df104a949ed523087a13b484f71bc9a351e50eec8b7
-size 181418766
+oid sha256:be6bf51c907265b4114c59ab98065830e9284f0e94df0759f7211b87ac90e8cd
+size 141679457

onnx/encoder_model.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3fdb4088122becb45da1c33e37bec35dfcdcd563b09aa30ac24db48b88d75369
-size 352812843
+oid sha256:31a05a14d514440e43746fdaaa8d4e8102c9543e53c5ae1111910af142041406
+size 352839389

onnx/encoder_model_quantized.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:736382d9bb09533d595c3c2d96e9f8716e9facfe10242b8ab4c8e802377665d3
-size 92289693
+oid sha256:969f5ac12974340386bf7a02ea6626003e5e2dee396ffc6ab0eec282bf55ba06
+size 92324809

quantize_config.json

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