Upgrade dependencies & replace deprecated modules with modern alternatives#136
Upgrade dependencies & replace deprecated modules with modern alternatives#136rohitgirdhar merged 8 commits intofacebookresearch:mainfrom
Conversation
|
Fixed the issues I ran into with the current installation procedure. Just had to modify a couple of the import statements to reflect their current implementation. One of them was torchvision.transforms.functional_tensor (became torchvision.transforms.functional) within the torchvision package and the other was the same in two of imagebind's files (timm.layers became timm.modules.layers). |
I couldn't find any references to functional_tensor in the code. Can you provide more details? I don't believe the timm v1.0.15 has The provided changes in this PR works with the versions defined in |
|
In 7436d3a, I removed some deps from the |
|
The current HEAD 3fcf5c9 does not work with modern dependencies. This PR makes ImageBind usable again for me (hopefully it will get merged), thanks @ahmedsaed! |
|
Any plans to merge it? The current version of ImageBind is broken anyway (for non-outdated versions of the dependencies) |
This PR upgrades some project dependencies and replaces deprecated modules with their recommended alternatives.
Upgraded dependencies:
6cdc9(latest at the time of making the PR)Replaced deprecated modules:
torchvision.transforms._transforms_video.NormalizeVideowithtorchvision.transforms.NormalizeI have tested these changes in this colab notebook using the example provided in the README.md.