HUB
Sign inSign up

armlimited/neural-graphics-model-gym

By armlimited

•Updated 4 months ago

A Python® toolkit for developing real-time Neural Graphics machine learning models.

Image
Machine learning & AI
Developer tools
2

521

armlimited/neural-graphics-model-gym repository overview

⁠Neural Graphics Model Gym

GitHubNSS Model CardNFRU Model Card

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:

  • Neural Super Sampling⁠ (NSS)
    • NSS allows for high-fidelity, real-time graphics in game engines. By feeding low-resolution frames, along with spatial and motion information, into a neural network we are able to construct high-resolution frames that suffer no loss in quality.
  • Neural Frame Rate Upscaling⁠ (NFRU)
    • NFRU allows for higher frame-rate real-time graphics in game engines. By feeding low-frame-rate frames, along with spatial and motion information, into a neural network we are able to construct intermediate frames that increase the output frame rate.

⁠Prerequisites

To run the Docker image the following are required:

  • NVIDIA® CUDA® capable GPU
  • NVIDIA Container Toolkit

⁠Setting flags for running the Docker image

⁠Access GPU in Docker container

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".

⁠Increase shared memory

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.

⁠Running the Docker 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>

⁠Running Neural Graphics Model Gym inside the container

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.

⁠Documentation

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.

⁠Software Bill of Materials (SBOM) and Licenses

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

⁠Trademarks and copyrights

  • Python® is a registered trademark of the Python Software Foundation.
  • Ubuntu® is a registered trademark of Canonical.
  • Docker and the Docker logo are trademarks or registered trademarks of Docker, Inc. in the United States and/or other countries. Docker, Inc. and other parties may also have trademark rights in other terms used herein.
  • NVIDIA and the NVIDIA logo are trademarks and/or registered trademarks of NVIDIA Corporation in the U.S. and other countries.
  • Vulkan is a registered trademark and the Vulkan SC logo is a trademark of the Khronos Group Inc.

Tag summary

Content type

Image

Digest

sha256:3474cd978…

Size

5.1 GB

Last updated

4 months ago

docker pull armlimited/neural-graphics-model-gym