Spaces:

finerobotics
/

annotation-studio

Running

Consolidate native hand and Flora Jobs in Annotation Studio; use the official dataset viewer

b00f3e3 verified 1 day ago

9.25 kB

FROM oven/bun:1.3.10 AS frontend
WORKDIR /build
COPY frontend/package.json frontend/bun.lock ./
RUN bun install --frozen-lockfile
COPY frontend/ ./
RUN bun run build

# Build Flora's dependencies on CPU without credentials or Flora model weights.
# Verified upstream PyTorch image digest, CUDA 12.8 / Torch 2.11.0.
FROM pytorch/pytorch:2.11.0-cuda12.8-cudnn9-devel@sha256:53ab3de62f6101d1e42f9be28623ab7a468a24c070d632f211ed576e30b6abd3 AS flora-build
ENV DEBIAN_FRONTEND=noninteractive TORCH_CUDA_ARCH_LIST=9.0 MAX_JOBS=4
RUN apt-get update && apt-get install -y --no-install-recommends python3-venv \
    && rm -rf /var/lib/apt/lists/* \
    && python -m venv --system-site-packages /opt/flora-venv
ENV PATH=/opt/flora-venv/bin:$PATH
WORKDIR /workspace/app
COPY worker-requirements.txt flora-requirements.txt flora-acceleration-requirements.txt ./
# Keep the reference worker/native dependency overlay order.
RUN python -m pip install --no-cache-dir -r worker-requirements.txt
RUN python -m pip install --no-cache-dir -r flora-requirements.txt
COPY build_causal_conv1d.py ./
ENV MAX_JOBS=1
RUN python build_causal_conv1d.py
RUN python -m pip install --no-cache-dir --no-build-isolation -r flora-acceleration-requirements.txt
RUN python -c "import causal_conv1d; from fla.ops.gated_delta_rule import chunk_gated_delta_rule, fused_recurrent_gated_delta_rule"
COPY *.py *.txt prompt-templates.json pipeline-source.tar.gz pipeline-version.json ./
RUN python bootstrap.py
COPY requirements.txt ./
RUN python -m venv --system-site-packages /opt/studio-venv
RUN /opt/studio-venv/bin/python -m pip install --no-cache-dir -r requirements.txt
RUN /opt/flora-venv/bin/python -c "from importlib.metadata import version; assert version('huggingface_hub') == '1.33.0'; assert version('transformers') == '5.8.1'"


# Camera operators arrive as verified wheels; this image needs no CUDA compiler.
FROM pytorch/pytorch:2.11.0-cuda12.8-cudnn9-runtime@sha256:eee11b3b3872a8c838e35ef48f08b2d5def2080902c7f666831310ca1a0ef2be AS hand-base
ENV DEBIAN_FRONTEND=noninteractive
RUN apt-get update && apt-get install -y --no-install-recommends git ffmpeg libglib2.0-0 python3-venv \
    && rm -rf /var/lib/apt/lists/*

# Isolate the controller, camera and hand dependencies to retain their CUDA ABIs.
FROM hand-base AS hand-build
ENV UV_PYTHON_INSTALL_DIR=/opt/uv-python
RUN python -m pip install --break-system-packages --no-cache-dir uv==0.12.11 \
    && uv python install 3.10.20 \
    && uv venv --python 3.10.20 /opt/droid \
    && uv venv --python 3.10.20 /opt/egohand \
    && python -m venv --system-site-packages /opt/hand-controller
WORKDIR /opt/hand-worker
COPY hand-runtime/camera-requirements.txt hand-runtime/egohand-requirements.txt hand-runtime/worker-requirements.txt ./
RUN uv pip install --python /opt/droid/bin/python torch==2.8.0 torchvision==0.23.0 --index-url https://download.pytorch.org/whl/cu128
RUN uv pip install --python /opt/droid/bin/python -r /opt/hand-worker/camera-requirements.txt \
    && uv pip install --python /opt/droid/bin/python torch-scatter==2.1.2+pt28cu128 --find-links https://data.pyg.org/whl/torch-2.8.0+cu128.html \
    && uv pip install --python /opt/droid/bin/python git+https://github.com/microsoft/MoGe.git@74fbce054ebed49800de42d0ad0e83495065719a --no-deps
RUN uv pip install --python /opt/egohand/bin/python -r /opt/hand-worker/egohand-requirements.txt \
    && /opt/hand-controller/bin/python -m pip install --no-cache-dir -r /opt/hand-worker/worker-requirements.txt
RUN git clone https://github.com/NVlabs/GEM-X.git /opt/GEM-X \
    && git -C /opt/GEM-X checkout --detach 20555befe2fbbe5b2015e1fff7ff214dc18ea9d6 \
    && uv pip install --python /opt/egohand/bin/python --no-deps -e /opt/GEM-X
COPY hand-runtime/camera-adapter-source.tar.gz hand-runtime/camera-native-source.tar.gz /tmp/
RUN mkdir -p /opt/camera-adapter /opt/hand-camera \
    && tar -xzf /tmp/camera-adapter-source.tar.gz -C /opt/camera-adapter \
    && tar -xzf /tmp/camera-native-source.tar.gz -C /opt/hand-camera
COPY hand-runtime/camera-wheels /opt/hand-worker/camera-wheels
COPY hand-runtime/verify_camera_wheels.py hand-runtime/image-version.json /opt/hand-worker/
RUN python /opt/hand-worker/verify_camera_wheels.py /opt/hand-worker/camera-wheels /opt/hand-worker/image-version.json \
    && uv pip install --python /opt/droid/bin/python --no-deps /opt/hand-worker/camera-wheels/wheels/*.whl
# The frozen camera wrapper expects these source, checkpoint and plan aliases.
RUN for source in /opt/hand-camera/release-195fps-verified/DROID-W/*; do \
      name=$(basename "$source"); \
      if [ ! -e "/opt/hand-camera/DROID-on/$name" ]; then ln -s "$source" "/opt/hand-camera/DROID-on/$name"; fi; \
    done \
    && mkdir -p /opt/hand-camera/moge-joint-vio/weights \
    && ln -s /workspace/hand-assets/camera/moge-small.pt /opt/hand-camera/moge-joint-vio/weights/small.pt \
    && ln -s /workspace/hand-assets/camera /opt/hand-camera/release-195fps-verified/DROID-W/pretrained \
    && ln -s /opt/hand-camera/release-195fps-verified/DROID-W/pretrained /opt/hand-camera/DROID-on/pretrained \
    && ln -s /workspace/hand-cache/camera-artifacts /opt/hand-camera/speed-round24/artifacts
RUN rm -rf /opt/hand-camera/release-195fps-verified/DROID-W/build \
    /opt/hand-camera/release-195fps-verified/DROID-W/thirdparty/lietorch/build /tmp/*.tar.gz

FROM hand-build AS hand-models
COPY hand-runtime/model-assets/ /model-assets/
COPY hand-runtime/model-sources.json hand-runtime/build_model_assets.py hand-runtime/egohand-checkpoint-template.zip /opt/hand-worker/
RUN /opt/egohand/bin/python /opt/hand-worker/build_model_assets.py /model-assets /opt/hand-worker/model-sources.json

FROM hand-base AS hand-runtime
COPY --from=hand-build /opt/uv-python /opt/uv-python
COPY --from=hand-build /opt/droid /opt/droid
COPY --from=hand-build /opt/egohand /opt/egohand
COPY --from=hand-build /opt/hand-controller /opt/hand-controller
COPY --from=hand-build /opt/GEM-X /opt/GEM-X
COPY --from=hand-build /opt/camera-adapter /opt/camera-adapter
COPY --from=hand-build /opt/hand-camera /opt/hand-camera
COPY hand-runtime/image-version.json hand-runtime/health.py hand-runtime/verify_runtime_imports.py /opt/hand-worker/
COPY hand-runtime/camera-startup-source.tar.gz /tmp/
COPY hand-runtime/camera-startup-sources.json hand-runtime/install_startup_sources.py /opt/hand-worker/
RUN python /opt/hand-worker/install_startup_sources.py /tmp/camera-startup-source.tar.gz /opt/hand-worker/camera-startup-sources.json /opt/hand-worker/image-version.json /opt/hand-camera \
    && rm /tmp/camera-startup-source.tar.gz
ENV PATH=/opt/hand-controller/bin:$PATH PYTHONUNBUFFERED=1 HF_HUB_DISABLE_TELEMETRY=1 PYTHONDONTWRITEBYTECODE=1
ENV HOME=/home/user XDG_CACHE_HOME=/home/user/.cache
ENV PYTHONPATH=/opt/camera-adapter/src GDP_TRT_UPDATE=1
ENV DROID_SITE=/opt/droid/lib/python3.10/site-packages EGO_SITE=/opt/egohand/lib/python3.10/site-packages
ENV DROID_LIBRARY_PATH=${DROID_SITE}/tensorrt_libs:${DROID_SITE}/nvidia/cu12/lib:${DROID_SITE}/nvidia/cudnn/lib:${DROID_SITE}/nvidia/cuda_runtime/lib:${DROID_SITE}/nvidia/cublas/lib:${DROID_SITE}/nvidia/cufft/lib:${DROID_SITE}/nvidia/curand/lib:${DROID_SITE}/nvidia/cuda_nvrtc/lib
ENV EGO_LIBRARY_PATH=${EGO_SITE}/tensorrt_libs:${EGO_SITE}/nvidia/cudnn/lib:${EGO_SITE}/nvidia/cu13/lib:${DROID_SITE}/nvidia/cuda_runtime/lib:${DROID_SITE}/nvidia/cublas/lib:${DROID_SITE}/nvidia/cufft/lib:${DROID_SITE}/nvidia/curand/lib:${DROID_SITE}/nvidia/cuda_nvrtc/lib
RUN if ! getent passwd 1000 >/dev/null; then useradd -m -u 1000 user; fi \
    && mkdir -p /workspace /home/user/.cache && chown -R 1000:1000 /workspace /home/user
USER 1000
WORKDIR /workspace
COPY --from=hand-models --chown=1000:1000 /model-assets /workspace/hand-assets
COPY hand-runtime/model-sources.json hand-runtime/build_model_assets.py /opt/hand-worker/
RUN python /opt/hand-worker/build_model_assets.py /workspace/hand-assets /opt/hand-worker/model-sources.json --verify-only
RUN python /opt/hand-worker/verify_runtime_imports.py all

# One public image supplies the Studio UI, Flora Jobs and native hand Jobs.
FROM hand-runtime AS studio
COPY --from=flora-build /opt/flora-venv /opt/flora-venv
COPY --from=flora-build /opt/studio-venv /opt/studio-venv
COPY --from=flora-build --chown=1000:1000 /workspace/source /workspace/source
WORKDIR /workspace/app
ENV PATH=/opt/flora-venv/bin:$PATH
ENV PYTHONPATH=/workspace/source/src:/workspace/app PYTHONUNBUFFERED=1 HF_HUB_DISABLE_TELEMETRY=1
ENV FLORA_EXECUTION=accelerated FLORA_BATCH_SIZE=8 FLORA_ENABLE_INFERENCE=1
COPY --chown=1000:1000 *.py *.txt *.json pipeline-source.tar.gz Dockerfile README.md THIRD_PARTY.md VISUALIZER-LICENSE TRANSFORMERS-LICENSE /workspace/app/
COPY --from=frontend --chown=1000:1000 /build/dist /workspace/app/frontend/dist
RUN python -c "import worker, torch, causal_conv1d; from fla.ops.gated_delta_rule import chunk_gated_delta_rule, fused_recurrent_gated_delta_rule; assert torch.__version__.split('+')[0] == '2.11.0'"
RUN PYTHONPATH=/opt/camera-adapter/src /opt/hand-controller/bin/python /opt/hand-worker/verify_runtime_imports.py all
USER 1000
EXPOSE 7860
CMD ["/opt/studio-venv/bin/python", "-m", "uvicorn", "server:app", "--host", "0.0.0.0", "--port", "7860", "--no-access-log"]

	FROM oven/bun:1.3.10 AS frontend
	WORKDIR /build
	COPY frontend/package.json frontend/bun.lock ./
	RUN bun install --frozen-lockfile
	COPY frontend/ ./
	RUN bun run build

	# Build Flora's dependencies on CPU without credentials or Flora model weights.
	# Verified upstream PyTorch image digest, CUDA 12.8 / Torch 2.11.0.
	FROM pytorch/pytorch:2.11.0-cuda12.8-cudnn9-devel@sha256:53ab3de62f6101d1e42f9be28623ab7a468a24c070d632f211ed576e30b6abd3 AS flora-build
	ENV DEBIAN_FRONTEND=noninteractive TORCH_CUDA_ARCH_LIST=9.0 MAX_JOBS=4
	RUN apt-get update && apt-get install -y --no-install-recommends python3-venv \
	&& rm -rf /var/lib/apt/lists/* \
	&& python -m venv --system-site-packages /opt/flora-venv
	ENV PATH=/opt/flora-venv/bin:$PATH
	WORKDIR /workspace/app
	COPY worker-requirements.txt flora-requirements.txt flora-acceleration-requirements.txt ./
	# Keep the reference worker/native dependency overlay order.
	RUN python -m pip install --no-cache-dir -r worker-requirements.txt
	RUN python -m pip install --no-cache-dir -r flora-requirements.txt
	COPY build_causal_conv1d.py ./
	ENV MAX_JOBS=1
	RUN python build_causal_conv1d.py
	RUN python -m pip install --no-cache-dir --no-build-isolation -r flora-acceleration-requirements.txt
	RUN python -c "import causal_conv1d; from fla.ops.gated_delta_rule import chunk_gated_delta_rule, fused_recurrent_gated_delta_rule"
	COPY .py .txt prompt-templates.json pipeline-source.tar.gz pipeline-version.json ./
	RUN python bootstrap.py
	COPY requirements.txt ./
	RUN python -m venv --system-site-packages /opt/studio-venv
	RUN /opt/studio-venv/bin/python -m pip install --no-cache-dir -r requirements.txt
	RUN /opt/flora-venv/bin/python -c "from importlib.metadata import version; assert version('huggingface_hub') == '1.33.0'; assert version('transformers') == '5.8.1'"


	# Camera operators arrive as verified wheels; this image needs no CUDA compiler.
	FROM pytorch/pytorch:2.11.0-cuda12.8-cudnn9-runtime@sha256:eee11b3b3872a8c838e35ef48f08b2d5def2080902c7f666831310ca1a0ef2be AS hand-base
	ENV DEBIAN_FRONTEND=noninteractive
	RUN apt-get update && apt-get install -y --no-install-recommends git ffmpeg libglib2.0-0 python3-venv \
	&& rm -rf /var/lib/apt/lists/*

	# Isolate the controller, camera and hand dependencies to retain their CUDA ABIs.
	FROM hand-base AS hand-build
	ENV UV_PYTHON_INSTALL_DIR=/opt/uv-python
	RUN python -m pip install --break-system-packages --no-cache-dir uv==0.12.11 \
	&& uv python install 3.10.20 \
	&& uv venv --python 3.10.20 /opt/droid \
	&& uv venv --python 3.10.20 /opt/egohand \
	&& python -m venv --system-site-packages /opt/hand-controller
	WORKDIR /opt/hand-worker
	COPY hand-runtime/camera-requirements.txt hand-runtime/egohand-requirements.txt hand-runtime/worker-requirements.txt ./
	RUN uv pip install --python /opt/droid/bin/python torch==2.8.0 torchvision==0.23.0 --index-url https://download.pytorch.org/whl/cu128
	RUN uv pip install --python /opt/droid/bin/python -r /opt/hand-worker/camera-requirements.txt \
	&& uv pip install --python /opt/droid/bin/python torch-scatter==2.1.2+pt28cu128 --find-links https://data.pyg.org/whl/torch-2.8.0+cu128.html \
	&& uv pip install --python /opt/droid/bin/python git+https://github.com/microsoft/MoGe.git@74fbce054ebed49800de42d0ad0e83495065719a --no-deps
	RUN uv pip install --python /opt/egohand/bin/python -r /opt/hand-worker/egohand-requirements.txt \
	&& /opt/hand-controller/bin/python -m pip install --no-cache-dir -r /opt/hand-worker/worker-requirements.txt
	RUN git clone https://github.com/NVlabs/GEM-X.git /opt/GEM-X \
	&& git -C /opt/GEM-X checkout --detach 20555befe2fbbe5b2015e1fff7ff214dc18ea9d6 \
	&& uv pip install --python /opt/egohand/bin/python --no-deps -e /opt/GEM-X
	COPY hand-runtime/camera-adapter-source.tar.gz hand-runtime/camera-native-source.tar.gz /tmp/
	RUN mkdir -p /opt/camera-adapter /opt/hand-camera \
	&& tar -xzf /tmp/camera-adapter-source.tar.gz -C /opt/camera-adapter \
	&& tar -xzf /tmp/camera-native-source.tar.gz -C /opt/hand-camera
	COPY hand-runtime/camera-wheels /opt/hand-worker/camera-wheels
	COPY hand-runtime/verify_camera_wheels.py hand-runtime/image-version.json /opt/hand-worker/
	RUN python /opt/hand-worker/verify_camera_wheels.py /opt/hand-worker/camera-wheels /opt/hand-worker/image-version.json \
	&& uv pip install --python /opt/droid/bin/python --no-deps /opt/hand-worker/camera-wheels/wheels/*.whl
	# The frozen camera wrapper expects these source, checkpoint and plan aliases.
	RUN for source in /opt/hand-camera/release-195fps-verified/DROID-W/*; do \
	name=$(basename "$source"); \
	if [ ! -e "/opt/hand-camera/DROID-on/$name" ]; then ln -s "$source" "/opt/hand-camera/DROID-on/$name"; fi; \
	done \
	&& mkdir -p /opt/hand-camera/moge-joint-vio/weights \
	&& ln -s /workspace/hand-assets/camera/moge-small.pt /opt/hand-camera/moge-joint-vio/weights/small.pt \
	&& ln -s /workspace/hand-assets/camera /opt/hand-camera/release-195fps-verified/DROID-W/pretrained \
	&& ln -s /opt/hand-camera/release-195fps-verified/DROID-W/pretrained /opt/hand-camera/DROID-on/pretrained \
	&& ln -s /workspace/hand-cache/camera-artifacts /opt/hand-camera/speed-round24/artifacts
	RUN rm -rf /opt/hand-camera/release-195fps-verified/DROID-W/build \
	/opt/hand-camera/release-195fps-verified/DROID-W/thirdparty/lietorch/build /tmp/*.tar.gz

	FROM hand-build AS hand-models
	COPY hand-runtime/model-assets/ /model-assets/
	COPY hand-runtime/model-sources.json hand-runtime/build_model_assets.py hand-runtime/egohand-checkpoint-template.zip /opt/hand-worker/
	RUN /opt/egohand/bin/python /opt/hand-worker/build_model_assets.py /model-assets /opt/hand-worker/model-sources.json

	FROM hand-base AS hand-runtime
	COPY --from=hand-build /opt/uv-python /opt/uv-python
	COPY --from=hand-build /opt/droid /opt/droid
	COPY --from=hand-build /opt/egohand /opt/egohand
	COPY --from=hand-build /opt/hand-controller /opt/hand-controller
	COPY --from=hand-build /opt/GEM-X /opt/GEM-X
	COPY --from=hand-build /opt/camera-adapter /opt/camera-adapter
	COPY --from=hand-build /opt/hand-camera /opt/hand-camera
	COPY hand-runtime/image-version.json hand-runtime/health.py hand-runtime/verify_runtime_imports.py /opt/hand-worker/
	COPY hand-runtime/camera-startup-source.tar.gz /tmp/
	COPY hand-runtime/camera-startup-sources.json hand-runtime/install_startup_sources.py /opt/hand-worker/
	RUN python /opt/hand-worker/install_startup_sources.py /tmp/camera-startup-source.tar.gz /opt/hand-worker/camera-startup-sources.json /opt/hand-worker/image-version.json /opt/hand-camera \
	&& rm /tmp/camera-startup-source.tar.gz
	ENV PATH=/opt/hand-controller/bin:$PATH PYTHONUNBUFFERED=1 HF_HUB_DISABLE_TELEMETRY=1 PYTHONDONTWRITEBYTECODE=1
	ENV HOME=/home/user XDG_CACHE_HOME=/home/user/.cache
	ENV PYTHONPATH=/opt/camera-adapter/src GDP_TRT_UPDATE=1
	ENV DROID_SITE=/opt/droid/lib/python3.10/site-packages EGO_SITE=/opt/egohand/lib/python3.10/site-packages
	ENV DROID_LIBRARY_PATH=${DROID_SITE}/tensorrt_libs:${DROID_SITE}/nvidia/cu12/lib:${DROID_SITE}/nvidia/cudnn/lib:${DROID_SITE}/nvidia/cuda_runtime/lib:${DROID_SITE}/nvidia/cublas/lib:${DROID_SITE}/nvidia/cufft/lib:${DROID_SITE}/nvidia/curand/lib:${DROID_SITE}/nvidia/cuda_nvrtc/lib
	ENV EGO_LIBRARY_PATH=${EGO_SITE}/tensorrt_libs:${EGO_SITE}/nvidia/cudnn/lib:${EGO_SITE}/nvidia/cu13/lib:${DROID_SITE}/nvidia/cuda_runtime/lib:${DROID_SITE}/nvidia/cublas/lib:${DROID_SITE}/nvidia/cufft/lib:${DROID_SITE}/nvidia/curand/lib:${DROID_SITE}/nvidia/cuda_nvrtc/lib
	RUN if ! getent passwd 1000 >/dev/null; then useradd -m -u 1000 user; fi \
	&& mkdir -p /workspace /home/user/.cache && chown -R 1000:1000 /workspace /home/user
	USER 1000
	WORKDIR /workspace
	COPY --from=hand-models --chown=1000:1000 /model-assets /workspace/hand-assets
	COPY hand-runtime/model-sources.json hand-runtime/build_model_assets.py /opt/hand-worker/
	RUN python /opt/hand-worker/build_model_assets.py /workspace/hand-assets /opt/hand-worker/model-sources.json --verify-only
	RUN python /opt/hand-worker/verify_runtime_imports.py all

	# One public image supplies the Studio UI, Flora Jobs and native hand Jobs.
	FROM hand-runtime AS studio
	COPY --from=flora-build /opt/flora-venv /opt/flora-venv
	COPY --from=flora-build /opt/studio-venv /opt/studio-venv
	COPY --from=flora-build --chown=1000:1000 /workspace/source /workspace/source
	WORKDIR /workspace/app
	ENV PATH=/opt/flora-venv/bin:$PATH
	ENV PYTHONPATH=/workspace/source/src:/workspace/app PYTHONUNBUFFERED=1 HF_HUB_DISABLE_TELEMETRY=1
	ENV FLORA_EXECUTION=accelerated FLORA_BATCH_SIZE=8 FLORA_ENABLE_INFERENCE=1
	COPY --chown=1000:1000 .py .txt *.json pipeline-source.tar.gz Dockerfile README.md THIRD_PARTY.md VISUALIZER-LICENSE TRANSFORMERS-LICENSE /workspace/app/
	COPY --from=frontend --chown=1000:1000 /build/dist /workspace/app/frontend/dist
	RUN python -c "import worker, torch, causal_conv1d; from fla.ops.gated_delta_rule import chunk_gated_delta_rule, fused_recurrent_gated_delta_rule; assert torch.__version__.split('+')[0] == '2.11.0'"
	RUN PYTHONPATH=/opt/camera-adapter/src /opt/hand-controller/bin/python /opt/hand-worker/verify_runtime_imports.py all
	USER 1000
	EXPOSE 7860
	CMD ["/opt/studio-venv/bin/python", "-m", "uvicorn", "server:app", "--host", "0.0.0.0", "--port", "7860", "--no-access-log"]