Update Dockerfile

Dockerfile

@@ -1,60 +1,48 @@
-# Base image
-FROM python:3.10-slim
+FROM ubuntu:22.04
 
-# Non-interactive environment
 ENV DEBIAN_FRONTEND=noninteractive
-ENV VENV_PATH=/BitNet/venv
-ENV PATH="$VENV_PATH/bin:$PATH"
 
-# Install system dependencies
+# Update & install system dependencies
 RUN apt-get update && apt-get install -y \
     wget \
     curl \
     git \
     cmake \
     build-essential \
-    clang \
-    libssl-dev \
-    libtinfo-dev \
+    python3 \
+    python3-pip \
+    python-is-python3 \
     ca-certificates \
     && rm -rf /var/lib/apt/lists/*
 
-# Clone BitNet repository
-RUN git clone --recursive https://github.com/microsoft/BitNet.git /BitNet
+# Install LLVM
+# NOTE: Uncomment this if you actually want to install LLVM.
+# RUN bash -c "$(wget -O - https://apt.llvm.org/llvm.sh)"
 
-# Set working directory
-WORKDIR /BitNet
+# Clone BitNet
+RUN git clone --recursive https://github.com/microsoft/BitNet.git
 
-# Copy local files (optional, if needed)
+# Copy everything from the local directory (where Dockerfile is) into the BitNet directory in the container
 COPY . /BitNet
 
-# Create Python virtual environment
-RUN python -m venv $VENV_PATH
-
-# Upgrade pip and install dependencies
-RUN pip install --upgrade pip
-RUN pip install -r requirements.txt
-RUN pip install huggingface_hub
+WORKDIR /BitNet
 
-# Download model from Hugging Face
-RUN huggingface-cli download HF1BitLLM/Llama3-8B-1.58-100B-tokens \
-    --local-dir models/Llama3-8B-1.58-100B-tokens
+# Install Python dependencies
+RUN pip install --upgrade pip && \
+    pip install -r requirements.txt
 
-# Setup environment / quantization
-RUN python setup_env.py -md ./models/Llama3-8B-1.58-100B-tokens -q i2_s
+# Hugging Face CLI setup
+RUN pip install huggingface_hub
 
-# Confirm files exist
-WORKDIR /BitNet/models/Llama3-8B-1.58-100B-tokens
-RUN ls -lah
+# Optional: HF token
+ARG HF_TOKEN
+RUN huggingface-cli login --token $HF_TOKEN || true
 
-# Reset working directory
-WORKDIR /BitNet
+# Download the model into ./models/
+RUN huggingface-cli download israellaguan/Falcon3-10B-Instruct-1.58bit-i2_s --local-dir models/
 
-# Expose port for inference server
-EXPOSE 7860
+# Run BitNet preprocessing script
+RUN python utils/codegen_tl1.py --model bitnet_b1_58-3B --BM 160,320,320 --BK 64,128,64 --bm 32,64,32
+RUN mkdir -p build && cd build && cmake .. && make -j$(nproc)
 
-# Default command to run inference server directly
-CMD ["python", "run_inference_server.py", \
-     "-m", "./models/Llama3-8B-1.58-100B-tokens/ggml-model-i2_s.gguf", \
-     "--host", "0.0.0.0", \
-     "--port", "7860"]
+CMD ["python", "app.py"]