I made a CPU only spiking neuron network lib that comes pretty close to PyTorch
A new library called NeuronGuard has been developed for spiking neural networks, showing promising results compared to PyTorch. It achieved 93.14% accuracy while training on 1,000,000 articles from the Wikimedia dataset in just 15.85 seconds on a standard Apple Silicon CPU. NeuronGuard's architecture allows for efficient training without the need for GPU acceleration, making it suitable for edge devices.
- ▪NeuronGuard is a neuromorphic spiking neural network library that operates efficiently on CPU hardware.
- ▪The model was trained on 1,000,000 articles and achieved 93.14% accuracy in a short training time.
- ▪It is significantly faster than PyTorch, with a training time of 15.85 seconds compared to 163.84 seconds for PyTorch with 10 epochs.
Opening excerpt (first ~120 words) tap to expand
etoxin / neuronguard-wikipedia-classifier like 0 Rust wikimedia/structured-wikipedia English neuromorphic spiking-neural-networks edge-ai green-ai License: apache-2.0 Model card Files Files and versions xet Community Copy to bucket new Neuromorphic Wikipedia Domain Classifier Performance Metrics (Standard Apple Silicon CPU) Comparison: NeuronGuard vs. PyTorch (1,000,000 Samples) Head-to-Head Results (Apple Silicon M-Series CPU/GPU)Architectural Trade-OffsTechnology Overview How to Load and Use in Python Open Source & Community Neuromorphic Wikipedia Domain Classifier This repository hosts the pre-trained vocabulary and synaptic weights for NeuronGuard, a proof of concept that uses cache-aligned Spiking Neural Network (SNN) and neuromorphic event engine written in Rust and exposed to…
Excerpt limited to ~120 words for fair-use compliance. The full article is at Huggingface.
Discussion
0 commentsMore from Huggingface
