Update README.md
Browse files
README.md
CHANGED
|
@@ -24,20 +24,8 @@ We release our model described in the paper
|
|
| 24 |
[Towards Better Instruction Following Language Models for Chinese](https://github.com/LianjiaTech/BELLE/blob/main/docs/Towards%20Better%20Instruction%20Following%20Language%20Models%20for%20Chinese.pdf)
|
| 25 |
|
| 26 |
This model achieves the best performance comparing other instruction-following models with a score of 0.762 on our evaluation set.
|
| 27 |
-
![](
|
| 28 |
-
|
| 29 |
-
<img src="https://huggingface.co/BelleGroup/belle-on-open-datasets/blob/main/main_results.png" width="450" height="auto">
|
| 30 |
-
</p>
|
| 31 |
-
|
| 32 |
-
## Training hyper-parameters
|
| 33 |
-
| Parameter | Value |
|
| 34 |
-
| ------ | ------ |
|
| 35 |
-
| Batch size | 16 |
|
| 36 |
-
| Learning rate | 5e-6 |
|
| 37 |
-
| Epochs | 3 |
|
| 38 |
-
|Weight_decay | 0.0 |
|
| 39 |
-
|Warmup_rate | 0.03 |
|
| 40 |
-
|LR_scheduler | cosine |
|
| 41 |
|
| 42 |
## Download, Convert & Check
|
| 43 |
1. After you git clone this model
|
|
|
|
| 24 |
[Towards Better Instruction Following Language Models for Chinese](https://github.com/LianjiaTech/BELLE/blob/main/docs/Towards%20Better%20Instruction%20Following%20Language%20Models%20for%20Chinese.pdf)
|
| 25 |
|
| 26 |
This model achieves the best performance comparing other instruction-following models with a score of 0.762 on our evaluation set.
|
| 27 |
+
|
| 28 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 29 |
|
| 30 |
## Download, Convert & Check
|
| 31 |
1. After you git clone this model
|