A Python® toolkit for developing real-time Neural Graphics machine learning models.
521
Neural Graphics Model Gym |
Components of this distribution have been gathered from many projects on the Internet, no security audit has been performed by Arm, and by choosing to use the distribution you thereby agree that you are taking the same risk as if you gathered the components from the many projects yourself, and that this distribution is being provided by Arm "AS IS" without warranty of any kind, express or implied, and that no liability is accepted by Arm.
Base Image Disclaimer Text
This image is provided by Arm and includes Arm-developed components on top of a publicly available base image (e.g., Ubuntu). Arm has not audited or tested the base OS or its third-party packages. For issues related to Arm's software, contact Arm. For base image concerns, contact the upstream provider.
Neural Graphics Model Gym is a Python® toolkit for developing real-time Neural Graphics machine learning models. This Docker® image provides dependencies, such as Python and system packages, required to run it.
With Neural Graphics Model Gym you can train, fine-tune and evaluate your Neural Graphics models. Neural Graphics Model Gym also enables you to perform quantization of your model before exporting it to a format compatible with ML extensions for Vulkan® - allowing you to run on the latest mobile devices.
Currently, we include the following Neural Graphics use cases:
To run the Docker image the following are required:
To access your GPU inside the Docker container, install the NVIDIA Container Toolkit then follow the instructions for Configuring Docker.
The GPU can then be accessed from the Docker container using the --gpus flag, either specifying --gpus all for access to all GPUs, or for specific GPUs pass the device numbers in a comma separated list e.g. --gpus "device=0,2".
To run training with the Docker image, the shared memory size must be increased from the default allocated to a container by using the --shm-size flag when running the container.
To run the container, specifying the shared memory size and access to all GPUs, use the following command:
docker run -it --shm-size=8gb --gpus all armlimited/neural-graphics-model-gym:<version_tag>
Neural Graphics Model Gym is available inside the container as source code in /home/neural-graphics-model-gym.
To use the CLI or the Python package, run:
cd /home/neural-graphics-model-gym
python -m pip install -U pip
pip install .
For usage and available commands, see /home/neural-graphics-model-gym/docs/usage.md.
For more details, including documentation on the commands which can be run from within the Docker container please refer to the README.md and usage.md for the corresponding version of the Neural Graphics Model Gym. These can also be found in the container under /home/neural-graphics-model-gym.
SBOM Disclaimer Text
THIS SOFTWARE BILL OF MATERIALS (“SBOM”) IS PROVIDED BY ARM LIMITED "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, AND NONINFRINGEMENT ARE DISCLAIMED. IN NO EVENT SHALL ARM LIMITED BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SBOM, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
A Software Bill of Materials (SBOM) and zip file of licenses covering all packages in the image, in addition to those from the base image, is attached to each image tag. These artifacts can be retrieved with oras, with the following commands:
oras discover registry-1.docker.io/armlimited/neural-graphics-model-gym:<tag>
oras pull registry-1.docker.io/armlimited/neural-graphics-model-gym@<sbom-digest>
oras pull registry-1.docker.io/armlimited/neural-graphics-model-gym@<licenses-digest>
E.g.
oras discover registry-1.docker.io/armlimited/neural-graphics-model-gym:0.3.0
registry-1.docker.io/armlimited/neural-graphics-model-gym@sha256:575bb4618dfb7096ee2224ebc843abef0ca7cab1c28b290276e7eca77f802118
├── application/spdx+json
│ └── sha256:3e3f30b19d5b243428dfe4bfb36de017559e21cffea9b5c3c9ffa3536c3b8488
└── application/zip
└── sha256:dc6da7e4bab51a427d557e2eb72e08ed0eafc14a2475824e0c5fc7f27e7a46ea
oras pull registry-1.docker.io/armlimited/neural-graphics-model-gym@sha256:3e3f30b19d5b243428dfe4bfb36de017559e21cffea9b5c3c9ffa3536c3b8488
✓ Pulled Neural-Graphics-Model-Gym_0.3.0_SPDX_SBOM.json 112/112 KB 100.00% 17ms
└─ sha256:3ad7825863eb9ad4d3e4712be1491337f1097db4936fed8c7f5bc2cf253e33f4
✓ Pulled application/vnd.oci.image.manifest.v1+json 775/775 B 100.00% 200µs
└─ sha256:3e3f30b19d5b243428dfe4bfb36de017559e21cffea9b5c3c9ffa3536c3b8488
Pulled [registry] registry-1.docker.io/armlimited/neural-graphics-model-gym@sha256:3e3f30b19d5b243428dfe4bfb36de017559e21cffea9b5c3c9ffa3536c3b8488
Digest: sha256:3e3f30b19d5b243428dfe4bfb36de017559e21cffea9b5c3c9ffa3536c3b8488
oras pull registry-1.docker.io/armlimited/neural-graphics-model-gym@sha256:dc6da7e4bab51a427d557e2eb72e08ed0eafc14a2475824e0c5fc7f27e7a46ea
✓ Pulled Neural-Graphics-Model-Gym_0.3.0_LICENSES.zip 343/343 KB 100.00% 49ms
└─ sha256:183577d458ed67a7b4fca7294b6b688f42a00d27dcf5f739425e2d1f935dfcd7
✓ Pulled application/vnd.oci.image.manifest.v1+json 767/767 B 100.00% 252µs
└─ sha256:dc6da7e4bab51a427d557e2eb72e08ed0eafc14a2475824e0c5fc7f27e7a46ea
Pulled [registry] registry-1.docker.io/armlimited/neural-graphics-model-gym@sha256:dc6da7e4bab51a427d557e2eb72e08ed0eafc14a2475824e0c5fc7f27e7a46ea
Digest: sha256:dc6da7e4bab51a427d557e2eb72e08ed0eafc14a2475824e0c5fc7f27e7a46ea
Content type
Image
Digest
sha256:3474cd978…
Size
5.1 GB
Last updated
4 months ago
docker pull armlimited/neural-graphics-model-gym