sami-ennedoui/stm32-datasheet-rag

Demo: optional per-request HF token (bring your own quota)

Graceful 503 on inference quota, 502 on non-convergence

Add agentic /agent/header endpoint (smolagents) and bump build

Rebuild bump to unstick scheduling

Deploy as Docker SDK: FastAPI + baked vector store + HTML demo

Switch to Docker SDK: drop gradio app.py

Pin huggingface-hub<1.0 (gradio 4.44 needs HfFolder)

Pin Python 3.12 (audioop removed in 3.13 breaks gradio/pydub)

Deploy STM32 datasheet RAG demo (Gradio)

initial commit