Dockerfile

# ============================================================
#  Dockerfile — Qwen3.5-0.8B CPU-Only API for HF Spaces
#  No GPU required. Port 7860.
# ============================================================

FROM python:3.11-slim

# ── System deps ──
RUN apt-get update && \
    apt-get install -y --no-install-recommends git && \
    rm -rf /var/lib/apt/lists/*

# ── Python deps (CPU-only torch — no CUDA bloat) ──
RUN pip install --no-cache-dir \
    torch --index-url https://download.pytorch.org/whl/cpu

RUN pip install --no-cache-dir \
    transformers \
    accelerate \
    fastapi \
    uvicorn \
    pydantic \
    huggingface_hub

# ── Pre-download model at build time (~1.8 GB baked into image) ──
ENV HF_HOME=/tmp/hf_cache
RUN python3 -c "\
from huggingface_hub import snapshot_download; \
snapshot_download('Qwen/Qwen2.5-0.5B-Instruct', cache_dir='/tmp/hf_cache')"

# ── Copy app ──
WORKDIR /app
COPY app.py .

EXPOSE 7860

CMD ["python3", "app.py"]