PERSON DIRECTORY
Tomasz Trzcinski
Tomasz Trzcinski appears in 1 indexed conversation across Latent Space. This directory brings every appearance, source, TL;DR, digest, and transcript into one searchable feed.
[NeurIPS Best Paper] 1000 Layer Networks for Self-Supervised RL — Kevin Wang et al, Princeton
Kevin WangIshaan JavaliMichał BortkiewiczTomasz TrzcinskiBenjamin Eysenbach
Self-supervised future-state classification, residual architecture, and extreme depth made 1,000-layer RL networks scale where naive widening or deepening failed, with gains emerging after roughly 50 million transitions.Depth improved parameter economics and larger-batch returns, while robotics, teacher-student distillation, and cheaper inference are prospective catalysts rather than proven deployments.