Spaces:
Running on Zero
Running on Zero
Commit History
-
Update app.py
Verified signatureca01502 -
Update app.py
Verified signature4985427
-
Update app.py
Verified signature0576de6 -
Update app.py
Verified signature376d91f -
Update app.py
Verified signature11fc23d
-
Hot-swap LoRA via Fast/Quality toggle, dynamic GPU duration, tqdm progress
Cache LoRA deltas in CPU fp32 so the 8-step LoRA can be added or subtracted in place between calls instead of being merged once at startup. Add a Fast/Quality radio that swaps num_steps and cfg_scale to the model card's recommended base settings, and dynamic @spaces.GPU duration (45s Fast / 95s Quality). Wrap the t2i and it2i sampling loops in tqdm so gr.Progress(track_tqdm=True) renders a per-step bar. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>4b58669
-
Add SenseNova-U1-8B-MoT demo with 8-step LoRA on ZeroGPU
Unified text-to-image and image editing UI, with the sensenova_u1 package vendored locally to register the NEO-Unify model architecture.84b5c54 -
Create app.py
Verified signature6e6cc43