Towards Improving Sequential Decision-Making in LLM Agents via Experience Memory

Jakub Rada, Viliam Lisý

Large language models have improved substantially on single-shot reasoning tasks, but their performance in sequential decision-making is less well understood. We study this on full…