anandpasupuleti/evaluate-pipeline

This Space now runs a FastAPI app inside Docker and uses local Hugging Face transformer models instead of Ollama.

Default models:

• —TEXT_MODEL_ID=google/gemma-4-E2B-it
• —OCR_MODEL_ID=zai-org/GLM-OCR

Local run:

pip install -r requirements.txt
uvicorn app:app --host 0.0.0.0 --port 7860

Docker run:

docker build -t evaluate-pipeline .
docker run -p 7860:7860 evaluate-pipeline

You can change the models with environment variables when deploying the Space if you want a smaller, faster, or more accurate checkpoint.