Update Dockerfile

Dockerfile

@@ -1,41 +1,36 @@
+# Use 3.11 to maintain compatibility with Torch 2.3.0
 FROM python:3.11-slim
 
-# 1. Install system dependencies
-# We keep build-essential for any C++ extensions needed by transformers/accelerate
+# 1. Install system dependencies for C++ builds (needed for sentencepiece)
 USER root
 RUN apt-get update && apt-get install -y \
     build-essential \
-    git \
     && rm -rf /var/lib/apt/lists/*
 
-# 2. Setup user
+# 2. Setup non-root user for Hugging Face
 RUN useradd -m -u 1000 user
 ENV HOME=/home/user \
     PATH=/home/user/.local/bin:$PATH \
     PYTHONUNBUFFERED=1 \
     PYTHONDONTWRITEBYTECODE=1 \
-    HF_HOME=/home/user/.cache/huggingface \
-    # Ensure torch doesn't try to use too many CPU threads before the GPU kicks in
-    OMP_NUM_THREADS=4
+    HF_HOME=/home/user/.cache/huggingface
 
 WORKDIR $HOME/app
 
-# 3. Create cache directories
-# 7B models are large (~15GB), so proper cache management is vital
+# 3. Initialize cache directories
 RUN mkdir -p $HF_HOME && chown -R user:user $HOME
 
 USER user
 
 # 4. Install Python requirements
-# Make sure 'spaces' and 'accelerate' are in your requirements.txt
 COPY --chown=user requirements.txt .
 RUN pip install --no-cache-dir --upgrade pip && \
     pip install --no-cache-dir -r requirements.txt
 
-# 5. Copy application
+# 5. Copy application code
 COPY --chown=user . .
 
-# Hugging Face ZeroGPU requires the app to run on port 7860
+# 6. Expose port 7860 for Hugging Face Spaces
 EXPOSE 7860
 
 CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]