LM-reasoning

Why we included this project

Anyone trying to get a handle on how reasoning works in large language models will find this index a good place to start. It is a hand-curated paper list organized by how reasoning gets elicited: fully supervised fine-tuning, prompting and in-context learning, hybrid methods, and evaluation and analysis. The chain-of-thought thread runs through it, from the original elicitation paper to zero-shot and program-aided variants, so you can see how the field moved rather than just collecting titles. Because it is tied to an accompanying survey paper, the selection reflects real editorial judgment about which works matter, and the maintainer accepts pull requests for missing papers. It is not software you deploy, but as a reading map for deciding which reasoning techniques to study or try, it saves a lot of scattered searching.