FROM nvidia/cuda:11.8.0-devel-ubuntu22.04 RUN apt update && apt install -y \ git \ build-essential \ libopenblas-dev \ wget \ python3-pip \ nodejs \ npm RUN useradd -m -u 1000 user USER user ENV HOME=/home/user \ PATH=/home/user/.local/bin:$PATH WORKDIR $HOME/app COPY --chown=user . $HOME/app RUN python3 -m pip install aphrodite-engine RUN wget -O part_a.gguf https://huggingface.co/NeverSleep/MiquMaid-v2-2x70B-DPO-GGUF/resolve/main/MiquMaid-v2-2x70B-DPO.Q4_K_M.gguf_part_aa?download=true RUN wget -O part_b.gguf https://huggingface.co/NeverSleep/MiquMaid-v2-2x70B-DPO-GGUF/resolve/main/MiquMaid-v2-2x70B-DPO.Q4_K_M.gguf_part_ab?download=true RUN cat part_a.gguf part_b.gguf > model.gguf RUN rm part_a.gguf part_b.gguf # Download config.json RUN wget -O config.json "https://huggingface.co/NeverSleep/MiquMaid-v2-2x70B-DPO/resolve/main/config.json?download=true" # Download special_tokens_map.json RUN wget -O special_tokens_map.json "https://huggingface.co/NeverSleep/MiquMaid-v2-2x70B-DPO/resolve/main/special_tokens_map.json?download=true" # Download tokenizer.json RUN wget -O tokenizer.json "https://huggingface.co/NeverSleep/MiquMaid-v2-2x70B-DPO/resolve/main/tokenizer.json?download=true" # Download tokenizer.model RUN wget -O tokenizer.model "https://huggingface.co/NeverSleep/MiquMaid-v2-2x70B-DPO/resolve/main/tokenizer.model?download=true" RUN wget -O tokenizer_config.json "https://huggingface.co/NeverSleep/MiquMaid-v2-2x70B-DPO/resolve/main/tokenizer_config.json?download=true" # Download tokenizer_config.json RUN wget -O tokenizer_config.json "https://huggingface.co/NeverSleep/MiquMaid-v2-2x70B-DPO/resolve/main/tokenizer_config.json?download=true" RUN wget -O gguf_to_torch.py https://raw.githubusercontent.com/PygmalionAI/aphrodite-engine/main/examples/gguf_to_torch.py RUN python3 -m pip install protobuf==3.20.* RUN python3 gguf_to_torch.py --input $HOME/app/model.gguf --output $HOME/app/ EXPOSE 7860 # Should expose on all network interfaces so 0.0.0.0 not only localhost CMD /bin/bash -c "/bin/python3 -m aphrodite.endpoints.openai.api_server $ENGINE_ARGS --port 7860 --host 0.0.0.0 --model $HOME/app/"