Update README.md
Browse files
README.md
CHANGED
|
@@ -6,8 +6,6 @@ datasets:
|
|
| 6 |
|
| 7 |
### Overview
|
| 8 |
|
| 9 |
-
__*NOTE: I found a bug that affected the 2.1 series of models, where the prompt was not being formatted properly during fine-tuning. It's been fixed here, and outputs are much better.*__
|
| 10 |
-
|
| 11 |
Another experimental model, using mostly sythetic data generated by [airoboros](https://github.com/jondurbin/airoboros)
|
| 12 |
|
| 13 |
Highlights:
|
|
@@ -17,6 +15,8 @@ Highlights:
|
|
| 17 |
- I re-generated all of the outputs in the dataset that had "Once upon a time" so they'd be less cliche - no guarantees that won't still happen, but in theory it may happen less.
|
| 18 |
- More multiple choice, better awareness, some alignment for normal use case but system-prompt overridable etc.
|
| 19 |
|
|
|
|
|
|
|
| 20 |
Breakdown of the training data:
|
| 21 |
|
| 22 |
| Count | Category |
|
|
|
|
| 6 |
|
| 7 |
### Overview
|
| 8 |
|
|
|
|
|
|
|
| 9 |
Another experimental model, using mostly sythetic data generated by [airoboros](https://github.com/jondurbin/airoboros)
|
| 10 |
|
| 11 |
Highlights:
|
|
|
|
| 15 |
- I re-generated all of the outputs in the dataset that had "Once upon a time" so they'd be less cliche - no guarantees that won't still happen, but in theory it may happen less.
|
| 16 |
- More multiple choice, better awareness, some alignment for normal use case but system-prompt overridable etc.
|
| 17 |
|
| 18 |
+
This model will very gladly spew profane and otherwise NSFW content, if asked, so just be careful about how you are prompting it!
|
| 19 |
+
|
| 20 |
Breakdown of the training data:
|
| 21 |
|
| 22 |
| Count | Category |
|