oBERT-12-upstream-pruned-unstructured-90-finetuned-qqp-v2

This model is obtained with The Optimal BERT Surgeon: Scalable and Accurate Second-Order Pruning for Large Language Models.

It corresponds to the model presented in the Table 2 - oBERT - QQP 90% (in the upcoming updated version of the paper).

Pruning method: oBERT upstream unstructured + sparse-transfer to downstream
Paper: https://arxiv.org/abs/2203.07259
Dataset: QQP
Sparsity: 90%
Number of layers: 12

The dev-set performance reported in the paper is averaged over four seeds, and we release the best model (marked with (*)):

| oBERT 90%     | acc   | F1    |
| ------------- | ----- | ----- |
| seed=42       | 90.94 | 87.79 |
| seed=3407     | 91.00 | 87.81 |
| seed=123      | 90.94 | 87.73 |
| seed=12345 (*)| 91.07 | 87.92 |
| ------------- | ----- | ----- |
| mean          | 90.99 | 87.81 |
| stdev         | 0.061 | 0.079 |

Code: https://github.com/neuralmagic/sparseml/tree/main/research/optimal_BERT_surgeon_oBERT

If you find the model useful, please consider citing our work.

Citation info

@article{kurtic2022optimal,
  title={The Optimal BERT Surgeon: Scalable and Accurate Second-Order Pruning for Large Language Models},
  author={Kurtic, Eldar and Campos, Daniel and Nguyen, Tuan and Frantar, Elias and Kurtz, Mark and Fineran, Benjamin and Goin, Michael and Alistarh, Dan},
  journal={arXiv preprint arXiv:2203.07259},
  year={2022}
}