Update README.md
Browse files
README.md
CHANGED
|
@@ -64,6 +64,13 @@ See [bagel](https://github.com/jondurbin/bagel) for additional details on the da
|
|
| 64 |
|
| 65 |
The DPO version will be available soon [here](https://huggingface.co/jondurbin/bagel-dpo-8b-v1.0)
|
| 66 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 67 |
### Data sources
|
| 68 |
|
| 69 |
There are many data sources used in the bagel models. See https://github.com/jondurbin/bagel for more information.
|
|
|
|
| 64 |
|
| 65 |
The DPO version will be available soon [here](https://huggingface.co/jondurbin/bagel-dpo-8b-v1.0)
|
| 66 |
|
| 67 |
+
Results look promising in comparison to mistral-7b-v0.2, e.g. MT-Bench:
|
| 68 |
+
|
| 69 |
+
| model | first turn | second turn | average |
|
| 70 |
+
| --- | --- | --- |
|
| 71 |
+
| bagel-8b-v1.0 | 7.64375 | 6.95 | 7.296875 |
|
| 72 |
+
| bagel-7b-v0.5 | 7.33125 | 6.8625 | 7.096875 |
|
| 73 |
+
|
| 74 |
### Data sources
|
| 75 |
|
| 76 |
There are many data sources used in the bagel models. See https://github.com/jondurbin/bagel for more information.
|