Update Dockerfile

Dockerfile

@@ -1,23 +1,20 @@
-# Using bullseye for better compatibility with build tools
 FROM python:3.10-bullseye
 
 WORKDIR /app
 
-# Install curl to download the model
+# Install only the essentials
 RUN apt-get update && apt-get install -y curl && rm -rf /var/lib/apt/lists/*
 
-# Fix: This is the magic line that skips the "Building wheel" phase
-# It pulls a pre-compiled CPU binary for llama-cpp-python
-RUN pip install --no-cache-dir \
-    fastapi \
-    uvicorn \
-    --extra-index-url https://abetlen.github.io/llama-cpp-python/whl/cpu \
-    llama-cpp-python
+# Force PIP to use CPU-only binaries and avoid the build-stage entirely
+ENV PIP_PREFER_BINARY=1
+RUN pip install --no-cache-dir fastapi uvicorn
+RUN pip install llama-cpp-python \
+    --extra-index-url https://abetlen.github.io/llama-cpp-python/whl/cpu
 
-# Download MPT-7B Q2
+# Download the model (MPT-7B Q2)
+# We do this AFTER the install so if the model fails, we don't have to reinstall tools
 RUN curl -L "https://huggingface.co/maddes8cht/mosaicml-mpt-7b-chat-gguf/resolve/main/mosaicml-mpt-7b-chat-Q2_K.gguf?download=true" -o mpt-7b-q2.gguf
 
-# Copy your backend code
 COPY app.py .
 
 EXPOSE 7860