Update Dockerfile

Dockerfile

@@ -1,10 +1,10 @@
-# Use lightweight Python base image
+# Use lightweight Python image
 FROM python:3.10-slim
- 
+
 # Set working directory
 WORKDIR /app
 
-# Install system dependencies
+# Install system dependencies (git needed for Hugging Face)
 RUN apt-get update && apt-get install -y git && rm -rf /var/lib/apt/lists/*
 
 # Copy requirements and install
@@ -14,18 +14,16 @@ RUN pip install --no-cache-dir -r requirements.txt
 # Copy application code
 COPY . .
 
-# ✅ Hugging Face cache directory
+# Set environment variables for cache dirs (Docker-friendly)
 ENV HF_HOME=/tmp
-# ENV TRANSFORMERS_CACHE=/tmp
+ENV TRANSFORMERS_CACHE=/tmp
+ENV TORCHINDUCTOR_CACHE_DIR=/tmp/torch_inductor_cache
 
-# Pre-download K2-Think model to avoid cold start
-# RUN python -c "from transformers import AutoTokenizer, AutoModelForCausalLM; \
-#     model_id='LLM360/K2-Think'; \
-#     AutoTokenizer.from_pretrained(model_id, cache_dir='/tmp'); \
-#     AutoModelForCausalLM.from_pretrained(model_id, cache_dir='/tmp')"
+# Create cache folder
+RUN mkdir -p /tmp/torch_inductor_cache
 
 # Expose port
 EXPOSE 7860
 
-# Command to run the FastAPI app
+# Start FastAPI
 CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "7860"]