Use editable installs for taming-transformers and CLIP (taming has no __init__.py)

PL 2.x: rank_zero_only moved from utilities.distributed to utilities

Shim mmcv.cnn.{constant,normal}_init for remote trust_remote_code modules

Add tensorboardX back (imported by depth/utils.py)

Replace removed torch._six.inf with math.inf (torch 2.x)

Move constant_init/normal_init imports to mmengine.model (mmcv 2.x split)

Use mmcv-lite (mmcv 2.x source build broken on setuptools >=81)

ZeroGPU app changes: @spaces.GPU, drop share=True, bump gradio 5.49.1

ZeroGPU: @spaces.GPU, drop cu113/torch 1.13 pin, modernize requirements, drop share=True

Upload app.py

Upload app.py

Upload app.py

Upload app.py

Upload app.py

Upload app.py

Upload app.py

Upload app.py

Upload app.py

Upload app.py

Upload app.py

Upload app.py

Upload app.py

Upload app.py

Upload app.py

Upload app.py

Upload app.py

Update requirements.txt

Delete depth/best_model_refcoco.pth

Delete depth/best_model_nyu.ckpt

Delete depth/best_model_kitti.ckpt

Update requirements.txt

Upload best_model_kitti.ckpt

Upload kitti_class_embeddings_my_captions.pth

init

init

init

init

Upload best_model_refcoco.pth

Upload mask_predictor.py

init

Upload best_model_refcoco.pth

init

Merge branch 'main' of hf.co:spaces/MykolaL/evp into main

init

Upload 2 files

init

init

init

init

init