You can deploy this model and start earning money today!

hazyresearch/
attn-360m

Name: attn-360m
Availability: InStock

text-generation-inference

endpoints_compatible

Views

No views yet

Model Card

This model is pretrained Attention (Llama architecture) model. The goal of this model is to provide a quality reference for the new Based architecture.

As a quality reference, we include a pretrained Mamba model provided here: https://huggingface.co/hazyresearch/mamba-360m, and a pretrained Based model provided here: https://huggingface.co/hazyresearch/based-360m

All three checkpoints are pretrained on 10Bn tokens of the Pile in the exact same data order using next token prediction.

Model Sources

The model implementation and training code that produced the model are provided here: https://github.com/HazyResearch/based

Uses

The purpose of this work is to evaluate the language modeling quality of a new efficient architecture, Based.

We include a series of benchmarks that you can use to evaluate quality:

Citation

Please consider citing this paper if you use our work:

@article{arora2024simple,
  title={Simple linear attention language models balance the recall-throughput tradeoff},
  author={Arora, Simran and Eyuboglu, Sabri and Zhang, Michael and Timalsina, Aman and Alberti, Silas and Zinsley, Dylan and Zou, James and Rudra, Atri and Ré, Christopher},
  journal={arXiv:2402.18668},
  year={2024}
}

Please reach out to simarora@stanford.edu, eyuboglu@stanford.edu, and mzhang20@stanford.edu with questions.

hazyresearch/attn-360m

Model Card

Model Sources

Uses

Citation

hazyresearch/
attn-360m