remove manual llama build

Dockerfile

@@ -1,42 +1,28 @@
 # Use an official Python runtime as a base image
-FROM python:3.10
+FROM python:3.10-slim
 
-# Set non-interactive for apt
-ENV DEBIAN_FRONTEND=noninteractive \
-    RUSTUP_HOME=/root/.rustup \
-    CARGO_HOME=/root/.cargo \
-    PATH=/root/.cargo/bin:$PATH \
-    TRANSFORMERS_CACHE=/app/cache \
-    HF_HOME=/app/cache
+# Set environment variables
+ENV TRANSFORMERS_CACHE=/app/cache \
+    HF_HOME=/app/cache \
+    DEBIAN_FRONTEND=noninteractive
 
 # Set working directory
 WORKDIR /app
 
-# Install system dependencies, Rust, and build tools
+# Install minimal system dependencies for running the app
 RUN apt-get update && apt-get install -y --no-install-recommends \
-      build-essential cmake git curl wget ninja-build libgomp1 ca-certificates \
-      gcc g++ libffi-dev libgcc-s1 libstdc++6 libopenblas-dev \
-    && rm -rf /var/lib/apt/lists/* \
-    && curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y \
-    && rustup default stable
+      libgomp1 \
+    && rm -rf /var/lib/apt/lists/*
 
 # Prepare cache directory
 RUN mkdir -p /app/cache && chmod -R 777 /app/cache
 
-# Copy and install Python requirements (excluding llama-cpp-python)
+# Copy requirements and install Python packages
 COPY requirements.txt .
-RUN sed -i '/llama-cpp-python/d' requirements.txt \
- && pip install --no-cache-dir -r requirements.txt
 
-
-# Clone & build llama-cpp-python (with its llama.cpp submodule)
-RUN git clone --recursive https://github.com/abetlen/llama-cpp-python.git /tmp/llama-cpp-python \
- && cd /tmp/llama-cpp-python \
- # ensure we have all submodules
- && git submodule update --init --recursive \
- # install from source
- && python -m pip install --no-cache-dir . \
- && rm -rf /tmp/llama-cpp-python
+# Install most packages from requirements.txt, then llama-cpp-python with binary-only
+RUN grep -v "llama-cpp-python" requirements.txt | pip install --no-cache-dir -r /dev/stdin \
+    && pip install --no-cache-dir --only-binary=all llama-cpp-python==0.3.14
 
 # Copy application code and data
 COPY app.py cv_embeddings.json cv_text.txt ./
@@ -45,4 +31,4 @@ COPY app.py cv_embeddings.json cv_text.txt ./
 EXPOSE 7860
 
 # Launch
-CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

requirements.txt

@@ -3,11 +3,22 @@ uvicorn==0.31.0
 sentence-transformers==3.1.1
 torch==2.4.1
 numpy==1.26.4
-llama-cpp-python==0.3.1
-huggingface_hub
-faiss-cpu
+huggingface_hub==0.30.1
+faiss-cpu==1.8.0
+psutil
+google-genai
+elevenlabs==1.1.3
+httpx==0.25.0
+llama-cpp-python==0.3.25.0
+uvicorn==0.31.0
+sentence-transformers==3.1.1
+torch==2.4.1
+numpy==1.26.4
+llama-cpp-python=0.3.14
+huggingface_hub==0.30.1
+faiss-cpu==1.8.0
 asyncio
 psutil
 google-genai
 elevenlabs
-httpx
+httpx==0.25.0