Update README.md
Browse files
README.md
CHANGED
|
@@ -6,6 +6,25 @@ license: apache-2.0
|
|
| 6 |
|
| 7 |
This repository contains the GGUF-v3 models (llama.cpp compatible) for **Chinese-LLaMA-2-13B**.
|
| 8 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
For Hugging Face version, please see: https://huggingface.co/hfl/chinese-llama-2-13b
|
| 10 |
|
| 11 |
Please refer to [https://github.com/ymcui/Chinese-LLaMA-Alpaca-2/](https://github.com/ymcui/Chinese-LLaMA-Alpaca-2/) for more details.
|
|
|
|
| 6 |
|
| 7 |
This repository contains the GGUF-v3 models (llama.cpp compatible) for **Chinese-LLaMA-2-13B**.
|
| 8 |
|
| 9 |
+
|
| 10 |
+
## Performance
|
| 11 |
+
|
| 12 |
+
Metric: PPL, lower is better
|
| 13 |
+
|
| 14 |
+
| Quant | original | imatrix (`-im`) |
|
| 15 |
+
|-----|------|------|
|
| 16 |
+
| Q2_K | 14.4701 +/- 0.26107 | 17.4275 +/- 0.31909 |
|
| 17 |
+
| Q3_K | 10.1620 +/- 0.18277 | 9.7486 +/- 0.17744 |
|
| 18 |
+
| Q4_K | 9.2735 +/- 0.16793 | 9.2734 +/- 0.16792 |
|
| 19 |
+
| Q5_K | 9.1767 +/- 0.16634 | 9.1594 +/- 0.16590 |
|
| 20 |
+
| Q6_K | 9.1326 +/- 0.16546 | 9.1478 +/- 0.16583 |
|
| 21 |
+
|
| 22 |
+
*The model with `-im` suffix is generated with important matrix, which has generally better performance (not always though).*
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
## Others
|
| 26 |
+
|
| 27 |
+
|
| 28 |
For Hugging Face version, please see: https://huggingface.co/hfl/chinese-llama-2-13b
|
| 29 |
|
| 30 |
Please refer to [https://github.com/ymcui/Chinese-LLaMA-Alpaca-2/](https://github.com/ymcui/Chinese-LLaMA-Alpaca-2/) for more details.
|