search

Efficient-Large-Model / VILA

VILA - a multi-image visual language model with training, inference and evaluation recipe, deployable from cloud to edge (Jetson Orin and laptops)
Apache License 2.0
878 stars 55 forks source link

fix: PR #40 other bug. #43

Closed SeanCraven314 closed 1 month ago

SeanCraven314 commented 1 month ago

Hi Again,

As mentioned in the #40, there is another problem when tensor ndim == 1 I'm not sure entirely what this is doing, but again I got the inference to work. It would be good if this could be reviewed.

Efficient-Large-Language-Model commented 1 month ago

Sorry I just rolled back your previous PR because it breaks other cases. Potentially it is fixed here. Could you rebase and we will review.

Efficient-Large-Language-Model commented 1 month ago

See my reply in https://github.com/Efficient-Large-Model/VILA/issues/39 Looks like this is not needed?

SeanCraven314 commented 1 month ago

See my reply in #39 Looks like this is not needed?

I agree we should close. I'm sorry for the breaking change.

tp-nan commented 1 month ago

i think tis PR may need reopen as #39 not really resoved