README.md

---
language: en
license: apache-2.0
tags:
- text-classfication
- int8
- Intel® Neural Compressor
- neural-compressor
- PostTrainingStatic
datasets: 
- mrpc
metrics:
- f1
---

# INT8 BERT base uncased finetuned MRPC

## Post-training static quantization

### PyTorch

This is an INT8  PyTorch model quantized with [huggingface/optimum-intel](https://github.com/huggingface/optimum-intel) through the usage of [Intel® Neural Compressor](https://github.com/intel/neural-compressor).

The original fp32 model comes from the fine-tuned model [Intel/bert-base-uncased-mrpc](https://huggingface.co./Intel/bert-base-uncased-mrpc).

The calibration dataloader is the train dataloader. The calibration sampling size is 1000.

The linear module **bert.encoder.layer.9.output.dense** falls back to fp32 to meet the 1% relative accuracy loss.

#### Test result

|   |INT8|FP32|
|---|:---:|:---:|
| **Accuracy (eval-f1)** |0.8959|0.9042|
| **Model size (MB)**  |119|418|

#### Load with Intel® Neural Compressor:

```python
from optimum.intel import INCModelForSequenceClassification

model_id = "Intel/bert-base-uncased-mrpc-int8-static"
int8_model = INCModelForSequenceClassification.from_pretrained(model_id)
```

### ONNX


This is an INT8 ONNX model quantized with [Intel® Neural Compressor](https://github.com/intel/neural-compressor).

The original fp32 model comes from the fine-tuned model [Intel/bert-base-uncased-mrpc](https://huggingface.co./Intel/bert-base-uncased-mrpc).

The calibration dataloader is the eval dataloader. The calibration sampling size is 100.

#### Test result

|   |INT8|FP32|
|---|:---:|:---:|
| **Accuracy (eval-f1)** |0.9021|0.9042|
| **Model size (MB)**  |236|418|


#### Load ONNX model:

```python
from optimum.onnxruntime import ORTModelForSequenceClassification
model = ORTModelForSequenceClassification.from_pretrained('Intel/bert-base-uncased-mrpc-int8-static')
```