Document automatic model path resolution in README

Resolve model paths automatically instead of a runAsContainer flag

Pin dependencies, drop dead route, rewrite README

Fix /cbow 500: use current Starlette TemplateResponse signature

Fix Space startup crash: drop device_map="auto" for CPU inference

Add GitHub Actions cron to keep HF Space awake

Pin transformers<5.0.0 to fix torch 2.6.0 incompatibility

Updated cbow UI

Trigger rebuild

UI Enhancements

fixing qwen model loading from HF dataset

moved both cbow and qwen to cache HF dataset

Handled reddit env info. Added back buttons.

Added two more games. Container runs locally

updated web pages

fix: write uploaded file to /tmp for HF compatibility

Fixed some format error in main.py

chore: trigger rebuild

The alive message is now visible. Switch back to original landing page.

revert back to non-lazy one

Fixed lazy loading caption_generator

fix: set app_port to 7860 to match Docker CMD

lazy loading

Add HEALTHCHECK and app_port

Debugging

Switch to uvicorn CMD for Hugging Face

need import of uvicorn

Add CMD to run main.py

Fix: update port to 8000 in main.py

Set Docker port to 8000

Add EXPOSE 8000 to Dockerfile

Fix: save CLIPProcessor in Docker build

Fix: preload bert tokenizer during Docker build

fix Dockerfile - load clip-vit-base-patch32 should be after pip install requirements.txt

Fix: preload CLIP model during Docker build

Fix: force cache to /tmp to avoid permission errors

HF needs all model downloads to a special read-write cache dir, also affects encoder.py

using /tmp, should be writable

removed /data, see if HF can use default internal cache

Need to create empty artifacts folder before wget

Dockerfile needs wget

Moved pth file to HF dataset

Resolve README.md conflict and merge with remote Hugging Face content

initial commit