Mosh Levy
Researcher in natural language processing.
Best podcasts with Mosh Levy
Ranked by the Snipd community
May 8, 2024 • 4min
arxiv preprint - Same Task, More Tokens: the Impact of Input Length on the Reasoning Performance of Large Language Models
Researchers Mosh Levy, Alon Jacoby, and Yoav Goldberg discuss how input length affects reasoning performance of Large Language Models (LLMs), revealing a drop in performance at shorter lengths. They explore limitations in traditional perplexity metrics, suggesting room for further research to enhance LLM reasoning abilities.
