WhirlwindAI/arithmetic-slm

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload app.py with huggingface_hub

Upload folder using huggingface_hub

Upload folder using huggingface_hub

Upload requirements.txt with huggingface_hub

Delete files custom_model/* with huggingface_hub

Upload folder using huggingface_hub

Improve output formatting: strip repeated expression, show clean answer

Load weights manually to avoid remote custom code shadowing local patches

Use local custom model code with transformers 5.x compatibility patches

Fix prompt template: use correct ⚠️ tags instead of warning emoji

Fix Gradio 6.x API: move theme/css to launch(), remove show_copy_button

Fix _finalize_model_loading as staticmethod with proper kwargs stripping

Patch init_weights and _finalize_model_loading to avoid passing extra kwargs

Universal tie_weights kwargs stripping for transformers 5.x

Add _tie_or_clone_weights for transformers 5.x compat

Unpin transformers (use 5.x with patch)

Patch tie_weights for transformers 5.x compatibility

Pin transformers<5 for custom_code compatibility

Upload folder using huggingface_hub

initial commit