LLM-KA-Cache-Compress / Dockerfile
cryptobiosis's picture
Fix compression-ratio display + switch to repetition-safe default prompt; align with Docker SDK manifest (Dockerfile + requirements.txt + README.md)
292780d verified
Raw
History Blame Contribute Delete
1.42 kB
# syntax=docker/dockerfile:1.6
#
# HuggingFace Space: kakeyalattice-demo (Docker SDK, Gradio inside)
#
# Build notes:
# - CPU-first so this runs on a free HF Space (no GPU required).
# - Pulls kakeyalattice + transformers + gradio from PyPI at build
# time, so the Space is self-contained and reproducible.
# - App runs Gradio on port 7860 (HF Space default).
FROM python:3.11-slim
# System deps (minimal — Gradio + torch-cpu don't need much)
RUN apt-get update && apt-get install -y --no-install-recommends \
git \
&& rm -rf /var/lib/apt/lists/*
# Create an unprivileged user (HF Spaces expect UID 1000)
RUN useradd -m -u 1000 user
USER user
ENV HOME=/home/user \
PATH=/home/user/.local/bin:$PATH \
PYTHONDONTWRITEBYTECODE=1 \
PYTHONUNBUFFERED=1 \
HF_HOME=/home/user/.cache/huggingface
WORKDIR $HOME/app
# Copy requirements first to maximise Docker layer cache
COPY --chown=user:user requirements.txt .
# Install CPU-only torch + gradio + our package from PyPI
# NOTE: --extra-index-url pulls CPU-only torch (smaller image, faster cold start on free tier).
RUN pip install --no-cache-dir --user --upgrade pip && \
pip install --no-cache-dir --user \
--extra-index-url https://download.pytorch.org/whl/cpu \
-r requirements.txt
# Copy the app
COPY --chown=user:user app.py README.md ./
# HF Space default port
EXPOSE 7860
CMD ["python", "app.py"]