Upload folder using huggingface_hub

config.json

@@ -1,5 +1,4 @@
 {
-  "_commit_hash": "dc68f91c06a1ba6f15268e5b9c13ae7a7c514084",
   "_name_or_path": "nlpconnect/vit-gpt2-image-captioning",
   "architectures": [
     "VisionEncoderDecoderModel"
@@ -92,7 +91,6 @@
     "top_p": 1.0,
     "torch_dtype": null,
     "torchscript": false,
-    "transformers_version": "4.29.2",
     "typical_p": 1.0,
     "use_bfloat16": false,
     "use_cache": true,
@@ -175,7 +173,6 @@
     "top_p": 1.0,
     "torch_dtype": null,
     "torchscript": false,
-    "transformers_version": "4.29.2",
     "typical_p": 1.0,
     "use_bfloat16": false
   },
@@ -184,5 +181,5 @@
   "model_type": "vision-encoder-decoder",
   "pad_token_id": 50256,
   "tie_word_embeddings": false,
-  "transformers_version": null
+  "transformers_version": "4.33.0.dev0"
 }

generation_config.json

@@ -4,5 +4,5 @@
   "decoder_start_token_id": 50256,
   "eos_token_id": 50256,
   "pad_token_id": 50256,
-  "transformers_version": "4.29.2"
+  "transformers_version": "4.33.0.dev0"
 }

onnx/decoder_model.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:722fc4a49f257eb69381633d21107308d01914601facf7f2143e0dedc4f4348a
-size 767520869
+oid sha256:14c3b794c998e17cad6f5e510504a6ce4e9607fb82e130907d39c9a6deb8f310
+size 613129810

onnx/decoder_model_merged.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:402ec6b9822f04e89fe325996d0aeb41d17ce239ce7828c85fb87093c10ebf6a
-size 768369266
+oid sha256:df4abab679e8fa8d5c2140dd1ce83d136e7d47bc08008c9d8b3860643c502776
+size 615025088

onnx/decoder_model_merged_quantized.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3c0f5f195147ca4087167acb7c9cd37b084bd12d1dea8792d23e0d0532630be7
-size 196406021
+oid sha256:1f3ec53b5fc3614c0b54ee786755a7ec3007841f57543f8e241499edfadfa98f
+size 158599996

onnx/decoder_model_quantized.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b06523e3e4832e2aeec9aa7c57785e9792c56871a0cf82c5171445932fc9b881
-size 195129787
+oid sha256:4cd3a527b1acd9893dec2ef8f02b5e759c0365e365976a2f44313c6095a8a256
+size 156278186

onnx/decoder_with_past_model.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6258fd83496643ae292beef05d3fe9e6968d425099d7353d34c678308d09bc98
-size 767518177
+oid sha256:926a856e44e926f6a865e8225009752615e7903a4a0b5641ff5ee9fe65cb9f14
+size 613127118

onnx/decoder_with_past_model_quantized.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:70af4a8110df4578662bb971ec10bcfdc045b7691ed7eb7e0ce51d4be8886ef8
-size 195121650
+oid sha256:a72d2c3557b4b5728e67eca3124682e0f3f1ed8aba1ae194556d03eaf3ccc693
+size 156270049

quantize_config.json

@@ -0,0 +1,125 @@
+{
+    "per_channel": true,
+    "reduce_range": true,
+    "per_model_config": {
+        "decoder_model": {
+            "op_types": [
+                "Add",
+                "Where",
+                "MatMul",
+                "Gather",
+                "Gemm",
+                "Reshape",
+                "Concat",
+                "Cast",
+                "Pow",
+                "Constant",
+                "ConstantOfShape",
+                "Slice",
+                "Mul",
+                "Div",
+                "Unsqueeze",
+                "Sub",
+                "Squeeze",
+                "Softmax",
+                "Split",
+                "Transpose",
+                "Shape",
+                "Tanh",
+                "Range",
+                "ReduceMean",
+                "Sqrt"
+            ],
+            "weight_type": "QInt8"
+        },
+        "encoder_model": {
+            "op_types": [
+                "Where",
+                "Add",
+                "Equal",
+                "MatMul",
+                "Gather",
+                "Conv",
+                "Reshape",
+                "Concat",
+                "Expand",
+                "Pow",
+                "Constant",
+                "ConstantOfShape",
+                "Erf",
+                "Slice",
+                "Mul",
+                "Div",
+                "Unsqueeze",
+                "Sub",
+                "Softmax",
+                "Transpose",
+                "Shape",
+                "ReduceMean",
+                "Sqrt"
+            ],
+            "weight_type": "QUInt8"
+        },
+        "decoder_model_merged": {
+            "op_types": [
+                "Add",
+                "Where",
+                "MatMul",
+                "Gather",
+                "Gemm",
+                "Reshape",
+                "Concat",
+                "Cast",
+                "Pow",
+                "Constant",
+                "ConstantOfShape",
+                "Slice",
+                "Mul",
+                "If",
+                "Div",
+                "Unsqueeze",
+                "Sub",
+                "Squeeze",
+                "Softmax",
+                "Split",
+                "Transpose",
+                "Shape",
+                "Tanh",
+                "Range",
+                "ReduceMean",
+                "Sqrt"
+            ],
+            "weight_type": "QInt8"
+        },
+        "decoder_with_past_model": {
+            "op_types": [
+                "Add",
+                "Where",
+                "MatMul",
+                "Gather",
+                "Gemm",
+                "Reshape",
+                "Concat",
+                "Cast",
+                "Pow",
+                "Constant",
+                "ConstantOfShape",
+                "Slice",
+                "Mul",
+                "Div",
+                "Unsqueeze",
+                "Squeeze",
+                "Sub",
+                "Softmax",
+                "Split",
+                "Transpose",
+                "Shape",
+                "Tanh",
+                "Range",
+                "ReduceMean",
+                "Sqrt"
+            ],
+            "weight_type": "QInt8"
+        }
+    }
+}

tokenizer_config.json

@@ -3,7 +3,15 @@
   "bos_token": "<|endoftext|>",
   "clean_up_tokenization_spaces": true,
   "eos_token": "<|endoftext|>",
+  "max_length": 32,
   "model_max_length": 1024,
+  "pad_to_multiple_of": null,
+  "pad_token": "<|endoftext|>",
+  "pad_token_type_id": 0,
+  "padding_side": "right",
+  "stride": 0,
   "tokenizer_class": "GPT2Tokenizer",
+  "truncation_side": "right",
+  "truncation_strategy": "longest_first",
   "unk_token": "<|endoftext|>"
 }