Hello, we noticed that SpikingJelly can be deployed to Loihi by converting a trained network to the lava framework. I would like to ask what the limitations are when designing the network structure, because we noticed that Loihi only supports 4096 input and output axons. If my convolution kernel is size (512,3,3), can't I convert to lava?
Hello, we noticed that SpikingJelly can be deployed to Loihi by converting a trained network to the lava framework. I would like to ask what the limitations are when designing the network structure, because we noticed that Loihi only supports 4096 input and output axons. If my convolution kernel is size (512,3,3), can't I convert to lava?