デバッグ出力強化

logの量を減らす

log追加

log 抑える

log出力変更戻す

log周りの修正

Fix model loading for repo ID and improve GPU detection

Load model directly from Hub to save storage space

Fix GPU detection by adding module-level GPU request and decorator on root endpoint

Add startup event to request GPU for ZeroGPU detection

Add token debugging and explicit login for gated model access

Fix model download logic for PyTorch models and improve error handling

Switch from llama-cpp-python to transformers for PyTorch model support

Fix ZeroGPU support, improve model loading and error handling

Move spaces.GPU() to module level for proper detection

Add GPU request for Hugging Face Space

Fix model path detection and update llama-cpp-python

Fix model path detection and update llama-cpp-python

Remove local model and add auto-download

Add FastAPI word tree server