Update README.md
Browse files
README.md
CHANGED
|
@@ -3,6 +3,9 @@ license: other
|
|
| 3 |
language:
|
| 4 |
- en
|
| 5 |
---
|
|
|
|
|
|
|
|
|
|
| 6 |
# Model Uses ChatML; Training details below.
|
| 7 |
### Single Epoch - 237 Step qlora test train at ebs-16 (bs 8 grad accumulation 2 lr of 1e-6 rank/alpha = 64): About 550 Conversation keys from each set using seed 69 to shuffle the sets.
|
| 8 |
|
|
|
|
| 3 |
language:
|
| 4 |
- en
|
| 5 |
---
|
| 6 |
+
|
| 7 |
+
# Using nightwing3 in the mix seems to have been a mistake. Redoing this train
|
| 8 |
+
|
| 9 |
# Model Uses ChatML; Training details below.
|
| 10 |
### Single Epoch - 237 Step qlora test train at ebs-16 (bs 8 grad accumulation 2 lr of 1e-6 rank/alpha = 64): About 550 Conversation keys from each set using seed 69 to shuffle the sets.
|
| 11 |
|