Orr Zohar comments

Results 16 comments of


                                            Orr Zohar

When doing pip install -v --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" ./, shows ModuleNotFoundError: No module named 'packaging'

I got the same error.

OwlVit gives different results compared to original colab version

Hi @MaslikovEgor, The demo didn't work before this fix as well (see https://github.com/huggingface/transformers/pull/20136). Try running coco evaluation with image conditioning before/after this fix, [email protected] increases from 6 to 37. This...

OwlVit gives different results compared to original colab version

@RRoundTable, the issues you are reporting seem to do with the text-conditioned evaluation. This means that the issues probably stem from the forward pass/post-processing. In your LVIS eval, did you...

OwlVit gives different results compared to original colab version

Hi @RRoundTable , I added a PR with the appropriate evaluation protocol https://github.com/huggingface/transformers/pull/23982 Best, Orr

OWOD data split doesn't work.

Even for the results in table 1 of your MS? Because you stated that you train on PASCAL VOC classes. Best Orr

access to llama-v1 weights (no response)

Hi,, It has been two and a half months and have yet to get access to the model weights. Will the weights ever be released? Best, Orr

too many requests for download video from YouTube

Same here

mm_use_im_start_end why turned on during instruction tuning and not pre-training [Question]

@haotian-liu Release notes are 404 Can you share the motivation for not using image start/end tokens?

[BUG] grad_norm and loss is nan when deepspeed==0.13.5 but ok with deepspeed==0.10.2

> Setting overlap_comm to False can avoid this problem. this solved the issue for me -- but what can we do to use overlap_comm? there is a significant slow down...

Video-LLaVa now available in the Transformers library!

@zucchini-nlp Does this support batch inferencing for faster evaluations?