Smnbrzq/muse-glimmer-pipecat-agent

pure fastapi: drop gradio to avoid double-bind

fix: pin gradio==6.28.0 to match builder

fix: gradio 5 for huggingface_hub compat

repurpose: Gemini Live phone relay

Create README.md

Create app.py

Create requirements.txt

Delete requirements.txt

Delete index.html

Delete app.py

Delete README.md

Delete .gitattributes

Make text-only example chips send immediately

Media example chips use public URLs from vl-test-suite

Add clickable example prompt chips

Use dtype= instead of deprecated torch_dtype

Simplify controls: use model generation defaults, keep reasoning strength only

Blog alignment: reasoning_strength template kwarg, video input support

Show reasoning in collapsible block; simplify header nav

Annotate chat return type so api outputs are inferred; filter reasoning preamble

Restyle chat UI to Meta AI dark theme with orb hero

Fix submit iterator: read status from event stream (no .on)

Serve homepage as HTMLResponse

Convert to gradio.Server backend + custom HTML/JS frontend

Switch to ZeroGPU xlarge (96GB) with 240s duration

Stream output tokens via TextIteratorStreamer

Fix examples format for additional_inputs (list of lists)

Switch to multimodal gr.ChatInterface with ZeroGPU large

Add torchvision dependency for image processor

Add Gradio app: HF-style logo maker + ZeroGPU multimodal chat

initial commit