Update app.py

Fix junk output: skip_special_tokens=False, lower temp, remove repetition_penalty, preserve </think> tag

Fix streaming: stop silent during thinking, yield answer after </think>

Remove type="messages" from ChatInterface - use compatible Gradio API

Rewrite with gr.ChatInterface to fix data incompatibility error

Pin gradio>=4.44.0 to support type="messages" in Chatbot

Fix: use Gradio messages format (type="messages") to resolve data incompatibility error

Lazy model loading - load model on first GPU request instead of startup

Fix indentation - clean app.py for Qwen3.5 AWQ ZeroGPU chat

Create requirements.txt

Add Qwen3.5-35B-A3B AWQ chat app with ZeroGPU support

initial commit

Duplicate from gradio-templates/chatbot