update README
Browse files
README.md
CHANGED
@@ -2,43 +2,42 @@
|
|
2 |
inference: false
|
3 |
---
|
4 |
|
5 |
-
#
|
6 |
|
7 |
-
|
8 |
|
9 |
-
|
10 |
|
11 |
-
- **Developed by:** [LMSYS](https://lmsys.org/)
|
12 |
-
- **Model type:** An auto-regressive language model based on the transformer architecture.
|
13 |
-
- **License:** Non-commercial license
|
14 |
-
- **Finetuned from model:** [LLaMA](https://arxiv.org/abs/2302.13971).
|
15 |
|
16 |
-
### Model Sources
|
17 |
|
18 |
-
- **Repository:** https://github.com/lm-sys/FastChat
|
19 |
-
- **Blog:** https://lmsys.org/blog/2023-03-30-vicuna/
|
20 |
-
- **Paper:** https://arxiv.org/abs/2306.05685
|
21 |
-
- **Demo:** https://chat.lmsys.org/
|
22 |
|
23 |
-
##
|
|
|
24 |
|
25 |
-
|
26 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
27 |
|
28 |
-
## How to Get Started with the Model
|
29 |
|
30 |
-
|
31 |
-
- APIs (OpenAI API, Huggingface API): https://github.com/lm-sys/FastChat/tree/main#api.
|
32 |
|
33 |
-
|
34 |
|
35 |
-
Vicuna
|
36 |
-
The training data is around 140K conversations collected from ShareGPT.com.
|
37 |
-
See more details in the "Training Details of Vicuna Models" section in the appendix of this [paper](https://arxiv.org/pdf/2306.05685.pdf).
|
38 |
|
39 |
-
|
|
|
|
|
|
|
40 |
|
41 |
-
|
42 |
|
43 |
-
|
44 |
-
|
|
|
|
|
|
2 |
inference: false
|
3 |
---
|
4 |
|
5 |
+
# vicuna-13b-v1.3-4bit-g128-awq
|
6 |
|
7 |
+
Vicuna is a chat assistant trained by [LMSYS](https://lmsys.org/). This is a 4-bit AWQ quantized Vicuna v1.3 model.
|
8 |
|
9 |
+
[AWQ](https://github.com/mit-han-lab/llm-awq) is an **efficient and accurate** low-bit weight quantization (INT3/4) for LLMs, supporting instruction-tuned models and multi-modal LMs.
|
10 |
|
|
|
|
|
|
|
|
|
11 |
|
|
|
12 |
|
|
|
|
|
|
|
|
|
13 |
|
14 |
+
## Reference
|
15 |
+
If you find AWQ useful or relevant to your research, please kindly cite the paper:
|
16 |
|
17 |
+
```bibtex
|
18 |
+
@article{lin2023awq,
|
19 |
+
title={AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration},
|
20 |
+
author={Lin, Ji and Tang, Jiaming and Tang, Haotian and Yang, Shang and Dang, Xingyu and Han, Song},
|
21 |
+
journal={arXiv},
|
22 |
+
year={2023}
|
23 |
+
}
|
24 |
+
```
|
25 |
|
|
|
26 |
|
27 |
+
## Vicuna Model Card
|
|
|
28 |
|
29 |
+
### Model Details
|
30 |
|
31 |
+
Vicuna is a chat assistant trained by fine-tuning LLaMA on user-shared conversations collected from ShareGPT.
|
|
|
|
|
32 |
|
33 |
+
- **Developed by:** [LMSYS](https://lmsys.org/)
|
34 |
+
- **Model type:** An auto-regressive language model based on the transformer architecture.
|
35 |
+
- **License:** Non-commercial license
|
36 |
+
- **Finetuned from model:** [LLaMA](https://arxiv.org/abs/2302.13971).
|
37 |
|
38 |
+
#### Model Sources
|
39 |
|
40 |
+
- **Repository:** https://github.com/lm-sys/FastChat
|
41 |
+
- **Blog:** https://lmsys.org/blog/2023-03-30-vicuna/
|
42 |
+
- **Paper:** https://arxiv.org/abs/2306.05685
|
43 |
+
- **Demo:** https://chat.lmsys.org/
|