Brandon Han comments

Results 13 comments of


                                            Brandon Han

RuntimeError in PyTorch 1.6

Hi, @Zhaoguanhua @zhangbin0917 Thanks for your sharing. However, do you have tried to fix the bug of DistSyncBatchnorm due to the same reason? I tried to modify the code just...

fashionpedia-api/baseline_predictor_demo.ipynb is going to be updated anytime soon?

Any update? It's been almost a year...

About structure details and attribute learning

Same questions here. Could you please share the item ID for each image? I cannot match the caption to the image.

ValueError: transformers.models.auto.spec is None

This may be useful https://github.com/PyTorchLightning/pytorch-lightning/issues/11524

comparison experiment

Same here. @yumingj Could you please release the pretrained predictor to predict the texture attributes of generated images? I need to do a fair comparison.

Hi, sorry for the late reply. I think the `output_dict` in the following forward function is what you need. https://github.com/BrandonHanx/mmf/blob/d63a31f83918ab60cd00cff3b72bd0e455ed1100/mmf/models/fashionvil/contrastive.py#L32-L57

Minimal working example

I assume this issue has been solved. Closed.

Running inference

Hi, sorry for the late reply. I don't recommend using `mmf.utils.inference`. You can get the features first and then use them to get the rank list offline.

Running inference

I assume this issue has been solved. Closed.

Question about your loss function

Hi, @lijing1996 Any idea about this? I face the same question. The loss function is too hard to understand for me...