README.md

metadata

language:
  - en
license: apache-2.0
tags:
  - text-classfication
  - int8
  - Intel® Neural Compressor
  - PostTrainingDynamic
  - onnx
datasets:
  - glue
metrics:
  - f1
model-index:
  - name: bart-large-mrpc-int8-dynamic
    results:
      - task:
          name: Text Classification
          type: text-classification
        dataset:
          name: GLUE MRPC
          type: glue
          args: mrpc
        metrics:
          - name: F1
            type: f1
            value: 0.9050847457627118

INT8 bart-large-mrpc

Post-training dynamic quantization

PyTorch

This is an INT8 PyTorch model quantized with huggingface/optimum-intel through the usage of Intel® Neural Compressor.

The original fp32 model comes from the fine-tuned model bart-large-mrpc.

Test result

	INT8	FP32
Accuracy (eval-f1)	0.9051	0.9120
Model size (MB)	547	1556.48

Load with optimum:

from optimum.intel import INCModelForSequenceClassification

model_id = "Intel/bart-large-mrpc-int8-dynamic"
int8_model = INCModelForSequenceClassification.from_pretrained(model_id)

ONNX

This is an INT8 ONNX model quantized with Intel® Neural Compressor.

The original fp32 model comes from the fine-tuned model bart-large-mrpc.

Test result

	INT8	FP32
Accuracy (eval-f1)	0.9236	0.9120
Model size (MB)	764	1555

Load ONNX model:

from optimum.onnxruntime import ORTModelForSequenceClassification
model = ORTModelForSequenceClassification.from_pretrained('Intel/bart-large-mrpc-int8-dynamic')