oBERT-12-upstream-pruned-unstructured-97

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

It corresponds to the upstream pruned model used as a starting point for sparse-transfer learning to downstream tasks presented in the Table 2 - oBERT - {SQuADv1, MNLI, QQP} - 97%.

Finetuned versions of this model for each downstream task are:

• SQuADv1: neuralmagic/oBERT-12-upstream-pruned-unstructured-97-finetuned-squadv1
• MNLI: neuralmagic/oBERT-12-upstream-pruned-unstructured-97-finetuned-mnli
• QQP: neuralmagic/oBERT-12-upstream-pruned-unstructured-97-finetuned-qqp

Pruning method: oBERT upstream unstructured
Paper: https://arxiv.org/abs/2203.07259
Dataset: BookCorpus and English Wikipedia
Sparsity: 97%
Number of layers: 12

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}
}