Neve_benchmarks
Deep learning experiments and benchmarks for ResNet-18 implemented in the custom Neve language
SITE CAPTUREBased on information published on the service's website. Selection and review policy
WHAT IT DOES
What does it do?
Neve_benchmarks is an experiment implementing deep learning in approximately 1,400 lines of code written in a custom programming language. It was tested on Linux with an RTX 4090 and provides Docker instructions for GPU library compatibility.
KEY FEATURES
Key features
- Provides a ResNet-18 experiment implemented in Neve
- Runs benchmarks using Frost and a network library
- Provides comparative benchmarks based on PyTorch
- Sets up a GPU library-compatible execution environment with Docker
- Supports Neve syntax highlighting for Vim and VSCode
- Provides experimental procedures using the CIFAR-10 dataset