Update Dockerfile

Dockerfile

@@ -1,39 +1,36 @@
-# Use official Python image with basic system utilities
 FROM python:3.11-slim
 
-# Install system dependencies
+# Install required dependencies
 RUN apt-get update && apt-get install -y \
-    wget \
-    build-essential \
+    git curl build-essential cmake \
     && rm -rf /var/lib/apt/lists/*
 
 # Set working directory
 WORKDIR /app
 
-# Set HF cache dir for transformers, chromadb, etc.
-ENV TRANSFORMERS_CACHE=/app/.cache \
-    HF_HOME=/app/.cache \
-    PIP_NO_CACHE_DIR=true
+# Create cache and vector DB dir
+RUN mkdir -p /app/.cache /app/vector_database && chmod -R 777 /app
 
-# Create required dirs with open permissions
-RUN mkdir -p /app/.cache /app/vector_database && chmod -R 777 /app/.cache /app/vector_database
+# Set environment vars
+ENV TRANSFORMERS_CACHE=/app/.cache \
+    HF_HOME=/app/.cache
 
-# Pre-download your quantized GGUF model from HF
-# (Replace with your exact file path if different)
-RUN wget https://huggingface.co/Kalpokoch/FinetunedQuantizedTinyLama/resolve/main/tinyllama_dop_q4_k_m.gguf -O /app/tinyllama_dop_q4_k_m.gguf
+# Preinstall llama-cpp-python with known working wheel version (skip build)
+RUN pip install llama-cpp-python==0.2.61
 
-# Install Python dependencies separately to leverage Docker caching
+# Install other dependencies
 COPY requirements.txt .
 RUN pip install -r requirements.txt
 
-# Install llama-cpp-python from wheel (avoid compilation)
-RUN pip install llama-cpp-python --prefer-binary
-
-# Copy the app code
+# Copy app code
 COPY . .
 
-# Expose port for FastAPI
+# Download GGUF model directly at container build time
+RUN curl -L -o /app/tinyllama_dop_q4_k_m.gguf \
+    https://huggingface.co/Kalpokoch/FinetunedQuantizedTinyLama/resolve/main/tinyllama_dop_q4_k_m.gguf
+
+# Expose app port
 EXPOSE 7860
 
-# Launch app
+# Run the app
 CMD ["uvicorn", "app.app:app", "--host", "0.0.0.0", "--port", "7860"]