search

arsedler9 / lfads-torch

A PyTorch implementation of Latent Factor Analysis via Dynamical Systems (LFADS) and AutoLFADS.
https://arxiv.org/abs/2309.01230
Apache License 2.0
81 stars 18 forks source link

data help #26

Open wwrlearner opened 1 week ago

wwrlearner commented 1 week ago

Hi, Andrew, Could you please provide me for the synth_data? I'm interested in the results of three synth_data in the paper supplymentary.

arsedler9 commented 1 week ago

Hey @wwrlearner, the lfads-torch paper didn't have a supplement or synthetic data, but maybe you're referring to this repo? You can find some synthetic datasets there.

wwrlearner commented 4 days ago

Thank you! But when I run the run_single.py with the lorenz data, I meet the error in the image. The data is constructed by the DataModule Configuration and Model Configuration. Could you please give me some special advice for the syn_data training? image

arsedler9 commented 1 day ago

Hi @wwrlearner, this part of the script is trying to plot RasterPlot.n_samples examples of spikes, inferred rates, and inferred inputs. In this case, RasterPlot.n_samples=3 (the default) and it looks like indexing is going out of bounds. Is your batch size smaller than 3? You can always turn off the RasterPlot callback by removing it from the list of callbacks in your config.