-
So exl2 does support llava-v1.5? |
Beta Was this translation helpful? Give feedback.
Answered by
turboderp
May 9, 2024
Replies: 1 comment
-
|
Yes, in a limited sense. You still need the vision tower from Transformers to produce the image feature embeddings, but now there's an interface for mixing the embeddings into the prompt. See #399 for more. It will hopefully become a full set of features at some point, maybe even dropping the Transformers dependency. Just different priorities right now. |
Beta Was this translation helpful? Give feedback.
0 replies
Answer selected by
Katehuuh
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Yes, in a limited sense. You still need the vision tower from Transformers to produce the image feature embeddings, but now there's an interface for mixing the embeddings into the prompt. See #399 for more.
It will hopefully become a full set of features at some point, maybe even dropping the Transformers dependency. Just different priorities right now.