---
license: apache-2.0
datasets:
- togethercomputer/RedPajama-Data-1T
language:
- en
pipeline_tag: text-generation
library_name: transformers
---

## BSL-160M

[paper](https://arxiv.org/abs/2410.07064) | [code](https://github.com/microsoft/LMOps/tree/main/data_selection)

**BSL-160M** is a 160M model with [Mistral](https://arxiv.org/abs/2310.06825) achitecture pre-trained from scratch on the CC split of [Redpajama](https://github.com/togethercomputer/RedPajama-Data).

**It is used as the baseline for [PDS-160M](https://huggingface.co/Data-Selection/PDS-160M).**

### Evaluation

PDS-selected data improves the performance of language models pre-trained from scratch and saves pre-training comptation. The improvement scales up to large model sizes.

<p align='left'>
    <img src="https://cdn-uploads.huggingface.co/production/uploads/624ac662102fcdff87be51b9/6undIr37d10qD73TDiPDK.png" width="600">
</p>

### Citation

```bibtex
@article{gu2024data,
  title={Data Selection via Optimal Control for Language Models},
  author={Gu, Yuxian and Dong, Li and Wang, Hongning and Hao, Yaru and Dong, Qingxiu and Wei, Furu and Huang, Minlie},
  journal={arXiv preprint arXiv:2410.07064},
  year={2024}
}
```