Glimpse Prediction Networks — Data And Code

This repository contains pretrained GPNs/GSNs, their activations, RDMs, and other such measures relevant to the analysis presented in the paper (https://arxiv.org/abs/2511.12715). These data are essential companions to the analysis scripts provided at https://github.com/KietzmannLab/GPN (wherein instructions are provided on how to use these data). These files (the most updated version) can also be found on https://doi.org/10.17605/OSF.IO/G29PD Briefly, ‘datasets’ contains test-515 COCO scenes and glimpse sequences.‘logs’ contain the net_params which are the weights of the training GPNs/GSNs, and perf_logs which are the corresponding performance metrics. ‘saved_actvs’ contains GPN internal activations and behavior on various measures. ‘rdms’ contains RDMs across layers for GPNs/GSNs and the alternative models. ‘GPN_vX’ consist of archived github repositories corresponding to the versions X. To stitch the full zips from the provided parts, run: cat saved_actvs.part_* > saved_actvs.zip cat rdms.part_* > rdms.zip