Upload 7 files

Dockerfile

@@ -1,32 +1,23 @@
 FROM python:3.11-slim
 
-# Install system dependencies
-# We keep build tools and musl for compatibility.
+# Install minimal build tools
 RUN apt-get update && apt-get install -y --no-install-recommends \
     build-essential \
     cmake \
-    libopenblas-dev \
-    libopenblas0 \
     pkg-config \
-    musl \
     && rm -rf /var/lib/apt/lists/*
 
-# Fix for musl-linked wheels
-RUN ln -s /lib/x86_64-linux-musl/libc.so /lib/libc.musl-x86_64.so.1 || true
-
 WORKDIR /app
 
 # Install Python deps
 COPY requirements.txt .
 RUN pip install --no-cache-dir -r requirements.txt
 
-# Use the latest version to support new GGUF architectures (Gemma 2/4/etc).
-# We remove version pinning to get the newest available build.
-RUN CMAKE_ARGS="-DGGML_BLAS=ON -DGGML_BLAS_VENDOR=OpenBLAS" \
-    pip install --no-cache-dir \
-    llama-cpp-python \
-    --prefer-binary \
-    --extra-index-url https://abetlen.github.io/llama-cpp-python/whl/cpu
+# Build the LATEST llama-cpp-python from source.
+# We disable BLAS (OpenBLAS/OpenCL) to significantly speed up compilation 
+# and avoid the 20-minute Hugging Face build timeout. 
+# Standard CPU optimization is still very fast for a 4B model.
+RUN CMAKE_ARGS="-DGGML_BLAS=OFF" pip install --no-cache-dir llama-cpp-python
 
 COPY . .
 

app.py

@@ -28,7 +28,9 @@ print(f"Model downloaded: {model_path}")
 print(f"File size: {os.path.getsize(model_path) / (1024*1024):.2f} MB")
 
 # Import llama_cpp after ensuring it's installed
+import llama_cpp
 from llama_cpp import Llama
+print(f"llama-cpp-python version: {llama_cpp.__version__}")
 
 # CPU thread count — match HF free tier (2 vCPUs)
 N_THREADS = int(os.environ.get("N_THREADS", "2"))