README.md

---
license: apache-2.0
datasets:
- deepmind/narrativeqa
metrics:
- f1
---

# LLoCO: Learning Long Contexts Offline
[**Paper**](https://arxiv.org/abs/2404.07979) | [**Code**](https://github.com/jeffreysijuntan/lloco)

Lloco-7b-nqa is the LoRA adaptor checkpoint finetuned from [AutoCompressor-Llama-2-7b-6k](https://huggingface.co/princeton-nlp/AutoCompressor-Llama-2-7b-6k/) and [meta-llama/Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) 
using the **LLoCO** method in [LLoCO: Learning Long Contexts Offline](https://arxiv.org/abs/2404.07979). It is instruction-tuned on the NarrativeQA training set.

**LLoCO** enables LLMs to process long-context efficiently by learning contexts offline through context compression and in-domain parameter-efficient finetuning with LoRA. This approach extends the effective context window of a 4k token LLaMA2-7B model to handle up to 128k tokens, while using 
30x fewer tokens and achieving up to 7.62x inference speed-up.

## Released LoRA Checkpoint
| Model           | LoRA Rank |   Dataset   |                         Link                           |
|:----------------|-----------|-------------|--------------------------------------------------------|
| Lloco-7b-quality|     8     | QuALITY     | [link](https://huggingface.co/xiuyul/Lloco-7b-quality/)|
| Lloco-7b-qasper |     8     | Qasper      | [link](https://huggingface.co/xiuyul/Lloco-7b-qasper/) |
| Lloco-7b-qmsum  |     8     | QMSum       | [link](https://huggingface.co/xiuyul/Lloco-7b-qmsum/)  |
| Lloco-7b-nqa    |     8     | NarrativeQA | [link](https://huggingface.co/xiuyul/Lloco-7b-nqa/)    |
| Lloco-7b-hqa    |     8     | HotpotQA    | [link](https://huggingface.co/xiuyul/Lloco-7b-hqa/)    |

## Citation 
If you find this project useful, please consider citing:

```
@article{tan2024lloco,
  title={LLoCO: Learning Long Contexts Offline},
  author={Tan, Sijun and Li, Xiuyu and Patil, Shishir and Wu, Ziyang and Zhang, Tianjun and Keutzer, Kurt and Gonzalez, Joseph E and Popa, Raluca Ada},
  journal={arXiv preprint arXiv:2404.07979},
  year={2024}
}
```

## Evaluation
Check out [LLoCO: Learning Long Contexts Offline](https://arxiv.org/abs/2404.07979) for evaluation results on various long-context tasks such as long document question answering and summarization.