rantyhhh/gemma4-e2b-q4-thinking

Gemma 4 E2B llama.cpp Server

Minimal Hugging Face Docker Space for serving lmstudio-community/gemma-4-E2B-it-GGUF with llama.cpp in text-only mode.

Configuration

Set the following Space secret before starting the container:

• —LLAMA_API_KEY: required API key enforced by llama-server

The container listens on port 7860 by default and respects the Hugging Face-provided PORT environment variable.

API usage

This Space exposes the llama.cpp HTTP server API.

Example health check:

curl -H "Authorization: Bearer $LLAMA_API_KEY" \
  https://YOUR_SPACE_URL/health

Example chat completion:

curl https://YOUR_SPACE_URL/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $LLAMA_API_KEY" \
  -d '{
    "model": "gemma-4-E2B-it-Q4_K_M.gguf",
    "messages": [
      {"role": "user", "content": "Explain why the sky looks blue."}
    ]
  }'

This deployment is intentionally text-only and does not include multimodal projection weights.