README.md

# oBERT-6-upstream-pretrained-dense

This model is obtained with [The Optimal BERT Surgeon: Scalable and Accurate Second-Order Pruning for Large Language Models](https://arxiv.org/abs/2203.07259).

It corresponds to 6 layers from `neuralmagic/oBERT-12-upstream-pretrained-dense`, pretrained with knowledge distillation. This model is used as a starting point for downstream finetuning and pruning runs presented in the `Table 3 - 6 Layers`.
The model can also be used for finetuning on any downstream task, as a starting point instead of the two times larger `bert-base-uncased` model.

Finetuned and pruned versions of this model on the SQuADv1 downstream task, as described in the paper:
- 0%: `neuralmagic/oBERT-6-downstream-dense-squadv1`
- 80% unstructured: `neuralmagic/oBERT-6-downstream-pruned-unstructured-80-squadv1`
- 80% block-4: `neuralmagic/oBERT-6-downstream-pruned-block4-80-squadv1`
- 90% unstructured: `neuralmagic/oBERT-6-downstream-pruned-unstructured-90-squadv1`
- 90% block-4: `neuralmagic/oBERT-6-downstream-pruned-block4-90-squadv1`

```
Training objective: masked language modeling (MLM) + knowledge distillation
Paper: https://arxiv.org/abs/2203.07259
Dataset: BookCorpus and English Wikipedia
Sparsity: 0%
Number of layers: 6
```

Code: _coming soon_

## BibTeX entry and citation info
```bibtex
@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}
}
```