Spaces:
Running on Zero
Running on Zero
Commit History
-
Fix ZeroGPU worker error + tokenizer AttributeError (requirements.txt)
Two issues currently break the Space: 1. --extra-index-url https://download.pytorch.org/whl/cpu installs a CPU-only torch, overwriting the CUDA build the ZeroGPU base image ships. ZeroGPU's spaces runtime calls CUDA-specific torch APIs that don't exist in the CPU build -> AttributeError on every GPU call. 2. Unpinned sentencepiece resolves to 0.2.2, which removed SentencePieceProcessor.SetEncodeExtraOptions. text_encoder.py calls it during tokenizer init, so model loading crashed on every inference. Fix: drop the CPU index and pin sentencepiece==0.2.0 (still exposes SetEncodeExtraOptions). Verified on a duplicate ZeroGPU Space: all tabs (PCA, depth/normals, supervised + zero-shot segmentation) run. Co-Authored-By: Claude <noreply@anthropic.com>38c1e6d
-
Add psutil dependency required by fast_pytorch_kmeans
4c17f43
-
Load encoder directly from its own repo
Verified signature0772f06
-
Revert "Fix extract_features_value_attention and add _get_all_blocks"
This reverts commit f55b5f107d6b313b89520b4fc260a34207adc461.4b9c38c -
Fix extract_features_value_attention and add _get_all_blocks
Verified signaturef55b5f1 -
Update app.py
Verified signaturebdd767e
-
Fix indentation in load_variant (token + from_pretrained) (#1)
- Fix indentation in load_variant (token + from_pretrained) (cabd053e959a7ba37b970c2e295f3410dca3d8fc) Co-authored-by: Gabriele Berton <gberton@users.noreply.huggingface.co>bacf2ba -
Update app.py
Verified signatureb99ab8d -
Fix: Use HF_TOKEN to access private models
f8fe515 -
TIPSv2 Feature Explorer
Explore TIPSv2 representations: PCA visualization, zero-shot segmentation, DPT depth/normals estimation, and supervised segmentation (ADE20K).3e1b5c2 -
initial commit
Verified signature191fbda