Upload folder using huggingface_hub

config.json

@@ -139,8 +139,7 @@
     50360,
     50361
   ],
-  "torch_dtype": "float32",
-  "transformers_version": "4.29.2",
+  "transformers_version": "4.30.2",
   "use_cache": true,
   "use_weighted_layer_sum": false,
   "vocab_size": 51864

generation_config.json

@@ -110,5 +110,5 @@
     50360,
     50361
   ],
-  "transformers_version": "4.29.2"
+  "transformers_version": "4.30.2"
 }

onnx/decoder_model_merged.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:407050fe2f75a5eb90b80fc7d758aabd4748c6cd24559e6a4581b6320bdaa864
+oid sha256:2a6bc26a22416cb5ca53f5db05a44fae39a56a38053bd172c60a77be35974151
 size 1926573

onnx/decoder_model_merged_quantized.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c402233f1d2a50cc7f2fb96b7af97a9f0d0b9ccd151df68116408125af6f33de
-size 517263427
+oid sha256:cd4c06f8bfbfb28f8a777a84618692ee05687ab598fef9dffb2fea11d3cd83c4
+size 515405408

onnx/decoder_model_quantized.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9813464d9702b3ac296213dadf3712e1af3f1c1824ce1edb535f5541b0cb191d
-size 515875253
+oid sha256:8fe4c89bbf88b377dd1d2862571d7611ce6888d8d1a055b61790050791d97deb
+size 514017234

onnx/decoder_with_past_model_quantized.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4ba30507ec53dbb96c6e55b6bc9664a3c27770b0c7fb50c807fb2c33af975d60
-size 464950674
+oid sha256:789bc74086187043db99dd1684f3d82ec61d15464dddc2c3daba21dec6e505d5
+size 463338655

onnx/encoder_model_quantized.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c11cab5779929dcfbfe465f640ec12a116bd41794a50ffc4815824e80aaef2d5
-size 314503795
+oid sha256:952aacc7ab4ff86afccc732a42c4d6c8babb7651653aeb9b0ff22adcc42dcc90
+size 313397176

quant_config.json

@@ -0,0 +1,89 @@
+{
+    "per_channel": false,
+    "reduce_range": false,
+    "per_model_config": {
+        "encoder_model": {
+            "op_types": [
+                "ReduceMean",
+                "Sqrt",
+                "Reshape",
+                "Mul",
+                "Transpose",
+                "Shape",
+                "Gather",
+                "Sub",
+                "Add",
+                "Pow",
+                "Conv",
+                "Constant",
+                "MatMul",
+                "Erf",
+                "Concat",
+                "Unsqueeze",
+                "Softmax",
+                "Div"
+            ],
+            "weight_type": "QUInt8"
+        },
+        "decoder_model": {
+            "op_types": [
+                "Shape",
+                "Range",
+                "Expand",
+                "Sub",
+                "Sqrt",
+                "Reshape",
+                "Gather",
+                "Transpose",
+                "Add",
+                "Pow",
+                "Constant",
+                "Concat",
+                "ReduceMean",
+                "Mul",
+                "Equal",
+                "MatMul",
+                "Less",
+                "Where",
+                "Slice",
+                "Squeeze",
+                "Cast",
+                "Erf",
+                "ConstantOfShape",
+                "Unsqueeze",
+                "Softmax",
+                "Div"
+            ],
+            "weight_type": "QInt8"
+        },
+        "decoder_model_merged": {
+            "op_types": [
+                "If"
+            ],
+            "weight_type": "QInt8"
+        },
+        "decoder_with_past_model": {
+            "op_types": [
+                "ReduceMean",
+                "Sqrt",
+                "Reshape",
+                "Gather",
+                "Mul",
+                "Shape",
+                "Transpose",
+                "Add",
+                "Pow",
+                "Constant",
+                "MatMul",
+                "Softmax",
+                "Div",
+                "Erf",
+                "Concat",
+                "Unsqueeze",
+                "Slice",
+                "Sub"
+            ],
+            "weight_type": "QInt8"
+        }
+    }
+}