Update README.md
Browse files
README.md
CHANGED
|
@@ -39,7 +39,7 @@ This is a very experimental model. I've never done a vision model before, so I r
|
|
| 39 |
|
| 40 |
This is, however, the same methodology as all the other abliterated-v3s, but applied to a vision model. The vision encoding layers are completely unmodified, but the *language*/chat model layers are abliterated.
|
| 41 |
|
| 42 |
-
Please feel free to file issues to report any weird/strange behaviour.
|
| 43 |
|
| 44 |
## Hang on, "abliterated"? Orthogonalization? Ablation? What is this?
|
| 45 |
|
|
|
|
| 39 |
|
| 40 |
This is, however, the same methodology as all the other abliterated-v3s, but applied to a vision model. The vision encoding layers are completely unmodified, but the *language*/chat model layers are abliterated.
|
| 41 |
|
| 42 |
+
Please feel free to file issues to report any weird/strange behaviour. Expect this model to see updates as needed.
|
| 43 |
|
| 44 |
## Hang on, "abliterated"? Orthogonalization? Ablation? What is this?
|
| 45 |
|