README.md

metadata

title: Llava Image Captioner
emoji: 👀
colorFrom: gray
colorTo: gray
sdk: gradio
sdk_version: 5.46.1
app_file: app.py
pinned: false

TinyLLaVA Hugging Face Space (CPU)

Instructions:

• This Space installs torch at runtime before importing TinyLLaVA to avoid build-time failures.
• Model used: bczhou/TinyLLaVA-1.5B (CPU-friendly). Change MODEL_PATH in app.py if you have more resources.
• Upload an image or enter an image URL, enter an optional prompt, then click Generate.

Use this repo in Hugging Face Spaces (CPU). It vendors a minimal tinyllava loader to avoid pip dependency conflicts. Startup: set the Space run command to: python start.py This will install CPU torch at runtime then start app.py.

Model: bczhou/TinyLLaVA-1.5B (can be changed in app.py). CPU inference may be slow.