The Garden of Forking Paths: Observing Dynamic Parameters Distribution in Large Language Models

arXiv:2403.08739v1 Announce Type: new
Abstract: A substantial gap persists in understanding the reasons behind the exceptional performance of the Transformer architecture in NLP. A particularly unexplored area involves the mechanistic description of how the distribution of parameters evolves over time during training. In this work we suggest that looking at the time evolution of the statistic distribution of model parameters, and specifically at bifurcation effects, can help understanding the model quality, potentially reducing training costs and evaluation efforts and empirically …

abstract architecture arxiv cond-mat.dis-nn cond-mat.stat-mech cs.ai cs.cl distribution dynamic gap language language models large language large language models nlp parameters performance training transformer transformer architecture type understanding work