Fetching metadata from the HF Docker repository...
-
Update README.md
-
Update requirements.txt
-
Update app.py
-
fix: move 3D dice script to head param for proper execution
-
fix: use Together provider via router for Qwen text model
-
fix: use full router URL as model param
-
fix: use huggingface_hub 0.26-0.x with explicit router URL
-
fix: migrate to new HF Inference API (router.huggingface.co)
-
fix: pin Python 3.10 for audioop compatibility
-
fix: pin Python 3.10 for audioop compatibility
-
changed to dice game
-
changed to dice game
-
changed to dice game
-
changed to dice game
-
changed to dice game
-
changed to dice game
-
changed to dice game
-
changed to dice game
-
Call Fal image edit endpoint directly
-
Strip provider suffix from image model
-
Switch app to image-to-image model
-
Add ZeroGPU decorator
-
Replace with simple text-to-image app
-
Switch Luna UI to English
-
Refine Luna UI readability
-
Fix Hugging Face inference routing
-
Remove @GPU from HTTP call to fix ZeroGPU firewall DNS errors
-
Use raw requests API call to decisively bypass Hugging Face Hub router
-
Bypass Inference Providers router and use classic chat_completion method
-
Switch to mistralai/Mistral-7B-Instruct-v0.2 as a foolproof HF inference provider model
-
Switch to Qwen2.5-7B-Instruct to avoid PRO tier 32B model restrictions on hf-inference
-
Hardcode MODEL_ID and remove response_format to fix inference API errors
-
Update default model to Qwen2.5-Coder-32B-Instruct for hf-inference provider compatibility
-
Update default model to Qwen2.5-32B-Instruct because Qwen3 is unsupported
-
Add optional spaces.GPU decorator for ZeroGPU compatibility
-
Fix gradio dependency conflict by removing from requirements.txt and updating sdk_version
-
new2
-
new
-
Set header to mini
-
Add spaces and @spaces.GPU decorator to fix startup issue
-
Pin space and remove tr language from metadata
-
Add Turkish language support to instructions and metadata
-
Disable Gradio SSR for Space
-
Add resume bullet improver Space
-
initial commit
