Commit
•
a1ab09d
1
Parent(s):
3463e31
Update README.md
Browse files
README.md
CHANGED
|
@@ -78,4 +78,8 @@ Hercules-2.0-Mistral-7B is fine-tuned from the following sources:
|
|
| 78 |
- No mixed precision was used, with the default dtype being bfloat16.
|
| 79 |
- Trained on 100,000 examples of Hercules-v2.0.
|
| 80 |
- No model parameters were frozen.
|
| 81 |
-
- This model was trained on OpenAI's ChatML prompt format.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 78 |
- No mixed precision was used, with the default dtype being bfloat16.
|
| 79 |
- Trained on 100,000 examples of Hercules-v2.0.
|
| 80 |
- No model parameters were frozen.
|
| 81 |
+
- This model was trained on OpenAI's ChatML prompt format.
|
| 82 |
+
|
| 83 |
+
# Updates
|
| 84 |
+
|
| 85 |
+
- 🔥 **February 3, 2024: This model scored an average of 62 on Open LLM Leaderboard, outperforming OpenHermes-2.5 and OpenChat-3.5.**
|