vedant33/supersede-demo-zerogpu
0
gradio 6: drop removed type= kwarg on Chatbot
chat-UI side-by-side demo (preview mode)
ZeroGPU: duration 110s, trim memory tokens to fit per-call cap
expose api_name=run for batch curation
fix _gen: template->tokenize->**inputs (apply_chat_template returns dict)
ZeroGPU fix: lazy-load model inside @spaces.GPU (no CUDA at import)
ZeroGPU: base vs trained live demo
initial commit