Update README.md
Browse files
README.md
CHANGED
|
@@ -19,12 +19,23 @@ language:
|
|
| 19 |
|
| 20 |
# Model Description
|
| 21 |
This model finetuned [trinity-v1](https://huggingface.co/jan-hq/trinity-v1) on [ultrafeedback_binarized_subset](jan-hq/ultrafeedback_binarized_subset)
|
| 22 |
-
(cleaned version)
|
| 23 |
|
| 24 |
More details about the training result [here](https://huggingface.co/jan-hq/trinity-v1-dpo-adapter).
|
| 25 |
|
| 26 |
# Prompt template
|
| 27 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 28 |
```
|
| 29 |
{system_message}
|
| 30 |
### Instruction:
|
|
|
|
| 19 |
|
| 20 |
# Model Description
|
| 21 |
This model finetuned [trinity-v1](https://huggingface.co/jan-hq/trinity-v1) on [ultrafeedback_binarized_subset](jan-hq/ultrafeedback_binarized_subset)
|
| 22 |
+
(cleaned version) for adapt ChatML prompt template.
|
| 23 |
|
| 24 |
More details about the training result [here](https://huggingface.co/jan-hq/trinity-v1-dpo-adapter).
|
| 25 |
|
| 26 |
# Prompt template
|
| 27 |
|
| 28 |
+
**ChatML**
|
| 29 |
+
|
| 30 |
+
```
|
| 31 |
+
<|im_start|>system
|
| 32 |
+
{system_message}<|im_end|>
|
| 33 |
+
<|im_start|>user
|
| 34 |
+
{prompt}<|im_end|>
|
| 35 |
+
<|im_start|>assistant
|
| 36 |
+
```
|
| 37 |
+
|
| 38 |
+
|
| 39 |
```
|
| 40 |
{system_message}
|
| 41 |
### Instruction:
|