fix: single-stage build, download model at runtime to avoid timeout

Dockerfile

@@ -1,48 +1,44 @@
-FROM python:3.11-slim AS builder
+FROM python:3.11-slim
 
-# Install build dependencies
+# Install build + runtime dependencies in one layer
 RUN apt-get update && apt-get install -y \
-    git cmake clang build-essential wget curl \
+    git cmake clang build-essential curl libgomp1 \
     && rm -rf /var/lib/apt/lists/*
 
-# Clone bitnet.cpp
-RUN git clone --recursive https://github.com/microsoft/BitNet.git /opt/BitNet
+# Clone bitnet.cpp (shallow clone for speed)
+RUN git clone --depth 1 --recursive https://github.com/microsoft/BitNet.git /opt/BitNet
 
-# Install BitNet Python deps (needed for setup_env.py)
+# Install BitNet Python deps
 RUN pip install --no-cache-dir -r /opt/BitNet/requirements.txt
 
-# Build bitnet.cpp with i2_s kernel support
-# setup_env.py does the cmake build and downloads the GGUF model
-RUN cd /opt/BitNet && python setup_env.py \
-    --hf-repo microsoft/bitnet-b1.58-2B-4T-gguf \
-    -q i2_s
-
-# ─── Runtime stage ───────────────────────────────────────────────────────────
-FROM python:3.11-slim
-
-RUN apt-get update && apt-get install -y \
-    libgomp1 curl \
-    && rm -rf /var/lib/apt/lists/*
+# Build bitnet.cpp WITHOUT downloading the model (just compile the binary)
+# We use cmake directly instead of setup_env.py to avoid the model download
+RUN cd /opt/BitNet && \
+    cmake -B build -DCMAKE_BUILD_TYPE=Release \
+        -DGGML_BITNET_ARM_TL1=OFF \
+        -DGGML_BITNET_X86_TL2=OFF && \
+    cmake --build build --config Release -j$(nproc) --target llama-server llama-cli
 
-# Create non-root user (required by HF Spaces)
+# Create non-root user
 RUN useradd -ms /bin/bash user
 WORKDIR /home/user/app
 
-# Copy built binaries and model from builder
-COPY --from=builder /opt/BitNet/build/bin /home/user/app/bin
-COPY --from=builder /opt/BitNet/models /home/user/app/models
+# Copy just the binaries we need
+RUN cp /opt/BitNet/build/bin/llama-server /home/user/app/ && \
+    cp /opt/BitNet/build/bin/llama-cli /home/user/app/ && \
+    rm -rf /opt/BitNet
 
-# Install Python app dependencies
+# Install Python app deps
 COPY requirements.txt .
 RUN pip install --no-cache-dir -r requirements.txt
 
 # Copy app files
-COPY app.py .
-COPY start.sh .
+COPY app.py start.sh ./
 RUN chmod +x start.sh
 
-# Fix permissions
-RUN chown -R user:user /home/user/app
+# Create model directory
+RUN mkdir -p /home/user/app/models && \
+    chown -R user:user /home/user/app
 
 USER user