---
pipeline_tag: text-generation
license: apache-2.0
language:
- zh
- en
---

# Model Card for Breeze-7B-Instruct-64k-v0_1 with ExLlamaV2 Quantization
Original model 原始模型: https://huggingface.co/MediaTek-Research/Breeze-7B-Instruct-64k-v0_1

This is a quantizated model from [MediaTek-Research/Breeze-7B-Instruct-64k-v0_1](https://huggingface.co/MediaTek-Research/Breeze-7B-Instruct-64k-v0_1) in exl2 format.

You are currently at the [main](/kennylam/Breeze-7B-Instruct-64k-v0_1-exl2/tree/main) branch, which provides only [measurement.json](measurement.json) used in the ExLlamaV2 quantization. Please take a look of your choices in following table of branches.

這裡是main branch, 只提供EvLlamaV2量化時所用到的[measurement.json](measurement.json)檔案。


[8.0bpw-h8](/kennylam/Breeze-7B-Instruct-64k-v0_1-exl2/tree/8.0bpw-h8) 8 bits per weight.

[6.0bpw-h6](/kennylam/Breeze-7B-Instruct-64k-v0_1-exl2/tree/6.0bpw-h6) 6 bits per weight.

[5.0bpw-h6](/kennylam/Breeze-7B-Instruct-64k-v0_1-exl2/tree/5.0bpw-h6) 5 bits per weight.

[4.0bpw-h6](/kennylam/Breeze-7B-Instruct-64k-v0_1-exl2/tree/4.0bpw-h6) 4 bits per weight.

[3.0bpw-h6](/kennylam/Breeze-7B-Instruct-64k-v0_1-exl2/tree/3.0bpw-h6) 3 bits per weight.


## Citation

```
@article{breeze7b2024,
  title={},
  author={},
  journal={arXiv},
  year={2024}
}
```