yuwenz commited on
Commit
c95a1bb
1 Parent(s): 262ea06

upload int8 onnx model

Browse files

Signed-off-by: yuwenzho <yuwen.zhou@intel.com>

Files changed (4) hide show
  1. README.md +37 -0
  2. config.json +37 -0
  3. model.onnx +3 -0
  4. tokenizer.json +0 -0
README.md CHANGED
@@ -1,3 +1,40 @@
1
  ---
 
2
  license: mit
 
 
 
 
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
+ language: en
3
  license: mit
4
+ tags:
5
+ - text-classfication
6
+ - int8
7
+ - Intel® Neural Compressor
8
+ - PostTrainingDynamic
9
+ - onnx
10
+ datasets:
11
+ - mrpc
12
+ metrics:
13
+ - f1
14
  ---
15
+
16
+ # INT8 roberta base finetuned MRPC
17
+
18
+ ## Post-training dynamic quantization
19
+
20
+ ### ONNX
21
+
22
+ This is an INT8 ONNX model quantized with [Intel® Neural Compressor](https://github.com/intel/neural-compressor).
23
+
24
+ The original fp32 model comes from the fine-tuned model [Intel/roberta-base-mrpc](https://huggingface.co/Intel/roberta-base-mrpc).
25
+
26
+ #### Test result
27
+
28
+ | |INT8|FP32|
29
+ |---|:---:|:---:|
30
+ | **Accuracy (eval-f1)** |0.9085|0.9138|
31
+ | **Model size (MB)** |122|476|
32
+
33
+
34
+ #### Load ONNX model:
35
+
36
+ ```python
37
+ from optimum.onnxruntime import ORTModelForSequenceClassification
38
+ model = ORTModelForSequenceClassification.from_pretrained('Intel/roberta-base-mrpc-int8-dynamic')
39
+ ```
40
+
config.json ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "roberta-base",
3
+ "architectures": [
4
+ "RobertaForSequenceClassification"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.1,
7
+ "bos_token_id": 0,
8
+ "classifier_dropout": null,
9
+ "eos_token_id": 2,
10
+ "finetuning_task": "mrpc",
11
+ "hidden_act": "gelu",
12
+ "hidden_dropout_prob": 0.1,
13
+ "hidden_size": 768,
14
+ "id2label": {
15
+ "0": "not_equivalent",
16
+ "1": "equivalent"
17
+ },
18
+ "initializer_range": 0.02,
19
+ "intermediate_size": 3072,
20
+ "label2id": {
21
+ "equivalent": 1,
22
+ "not_equivalent": 0
23
+ },
24
+ "layer_norm_eps": 1e-05,
25
+ "max_position_embeddings": 514,
26
+ "model_type": "roberta",
27
+ "num_attention_heads": 12,
28
+ "num_hidden_layers": 12,
29
+ "pad_token_id": 1,
30
+ "position_embedding_type": "absolute",
31
+ "problem_type": "single_label_classification",
32
+ "torch_dtype": "float32",
33
+ "transformers_version": "4.18.0",
34
+ "type_vocab_size": 1,
35
+ "use_cache": true,
36
+ "vocab_size": 50265
37
+ }
model.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4385f8989ef927321b1688a9994c5cb50b72ed4fe25ac256ea1be9ff497755ce
3
+ size 127041708
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff