hitorilabs
commited on
Commit
•
6ee341c
1
Parent(s):
9d99332
Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,9 @@
|
|
1 |
---
|
2 |
license: llama2
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: llama2
|
3 |
---
|
4 |
+
|
5 |
+
Trained using TRL, it didn't fit properly on my 3090 without significantly dropping batch size and applying 4-bit quantization.
|
6 |
+
|
7 |
+
It didn't exactly converge.
|
8 |
+
|
9 |
+
![training_run.png](https://cdn-uploads.huggingface.co/production/uploads/64075c834dc5f2846c96bc98/b-Tn5IDcRubZp_AyfLNg7.png)
|