https://github.com/ridgerchu/SpikeGPThttps://arxiv.org/abs/2302.1...

https://github.com/ridgerchu/SpikeGPT

https://arxiv.org/abs/2302.13939

Our preliminary experiments show that SpikeGPT remains competitive with non-spiking models on tested benchmarks, while maintaining 5x less energy consumption when processed on neuromorphic hardware that can leverage sparse, event-driven activations.



has anyone even considered what happens with SpikeGPT running on neuromorphic hard that is capable of learning on chip?


can you teach the model new concepts through conversation?

isn't this pretty exciting to think about


@shareman what do you think about this?

does running SpikeGPT in the cloud in neuromorphic hardware solve any issues with cost of training and inference?