Matches in SemOpenAlex for { <https://semopenalex.org/work/W4387074712> ?p ?o ?g. }
Showing items 1 to 71 of
71
with 100 items per page.
- W4387074712 abstract "Large language models (LLMs) can improve their accuracy on various tasks through iteratively refining and revising their output based on feedback. We observe that these revisions can introduce errors, in which case it is better to roll back to a previous result. Further, revisions are typically homogeneous: they use the same reasoning method that produced the initial answer, which may not correct errors. To enable exploration in this space, we present SCREWS, a modular framework for reasoning with revisions. It is comprised of three main modules: Sampling, Conditional Resampling, and Selection, each consisting of sub-modules that can be hand-selected per task. We show that SCREWS not only unifies several previous approaches under a common framework, but also reveals several novel strategies for identifying improved reasoning chains. We evaluate our framework with state-of-the-art LLMs (ChatGPT and GPT-4) on a diverse set of reasoning tasks and uncover useful new reasoning strategies for each: arithmetic word problems, multi-hop question answering, and code debugging. Heterogeneous revision strategies prove to be important, as does selection between original and revised candidates." @default.
- W4387074712 created "2023-09-27" @default.
- W4387074712 creator A5002337174 @default.
- W4387074712 creator A5015256182 @default.
- W4387074712 creator A5018805044 @default.
- W4387074712 creator A5031480753 @default.
- W4387074712 creator A5052467896 @default.
- W4387074712 creator A5075825791 @default.
- W4387074712 date "2023-09-20" @default.
- W4387074712 modified "2023-09-28" @default.
- W4387074712 title "SCREWS: A Modular Framework for Reasoning with Revisions" @default.
- W4387074712 doi "https://doi.org/10.48550/arxiv.2309.13075" @default.
- W4387074712 hasPublicationYear "2023" @default.
- W4387074712 type Work @default.
- W4387074712 citedByCount "0" @default.
- W4387074712 crossrefType "posted-content" @default.
- W4387074712 hasAuthorship W4387074712A5002337174 @default.
- W4387074712 hasAuthorship W4387074712A5015256182 @default.
- W4387074712 hasAuthorship W4387074712A5018805044 @default.
- W4387074712 hasAuthorship W4387074712A5031480753 @default.
- W4387074712 hasAuthorship W4387074712A5052467896 @default.
- W4387074712 hasAuthorship W4387074712A5075825791 @default.
- W4387074712 hasBestOaLocation W43870747121 @default.
- W4387074712 hasConcept C101468663 @default.
- W4387074712 hasConcept C114614502 @default.
- W4387074712 hasConcept C119857082 @default.
- W4387074712 hasConcept C127413603 @default.
- W4387074712 hasConcept C154945302 @default.
- W4387074712 hasConcept C168065819 @default.
- W4387074712 hasConcept C177264268 @default.
- W4387074712 hasConcept C199360897 @default.
- W4387074712 hasConcept C20162079 @default.
- W4387074712 hasConcept C201995342 @default.
- W4387074712 hasConcept C204321447 @default.
- W4387074712 hasConcept C2780451532 @default.
- W4387074712 hasConcept C33923547 @default.
- W4387074712 hasConcept C41008148 @default.
- W4387074712 hasConcept C66882249 @default.
- W4387074712 hasConcept C81917197 @default.
- W4387074712 hasConceptScore W4387074712C101468663 @default.
- W4387074712 hasConceptScore W4387074712C114614502 @default.
- W4387074712 hasConceptScore W4387074712C119857082 @default.
- W4387074712 hasConceptScore W4387074712C127413603 @default.
- W4387074712 hasConceptScore W4387074712C154945302 @default.
- W4387074712 hasConceptScore W4387074712C168065819 @default.
- W4387074712 hasConceptScore W4387074712C177264268 @default.
- W4387074712 hasConceptScore W4387074712C199360897 @default.
- W4387074712 hasConceptScore W4387074712C20162079 @default.
- W4387074712 hasConceptScore W4387074712C201995342 @default.
- W4387074712 hasConceptScore W4387074712C204321447 @default.
- W4387074712 hasConceptScore W4387074712C2780451532 @default.
- W4387074712 hasConceptScore W4387074712C33923547 @default.
- W4387074712 hasConceptScore W4387074712C41008148 @default.
- W4387074712 hasConceptScore W4387074712C66882249 @default.
- W4387074712 hasConceptScore W4387074712C81917197 @default.
- W4387074712 hasLocation W43870747121 @default.
- W4387074712 hasOpenAccess W4387074712 @default.
- W4387074712 hasPrimaryLocation W43870747121 @default.
- W4387074712 hasRelatedWork W1483845062 @default.
- W4387074712 hasRelatedWork W1493324536 @default.
- W4387074712 hasRelatedWork W1522854984 @default.
- W4387074712 hasRelatedWork W1578053891 @default.
- W4387074712 hasRelatedWork W1601811574 @default.
- W4387074712 hasRelatedWork W1602801198 @default.
- W4387074712 hasRelatedWork W1987935534 @default.
- W4387074712 hasRelatedWork W2120071210 @default.
- W4387074712 hasRelatedWork W2321302561 @default.
- W4387074712 hasRelatedWork W97732546 @default.
- W4387074712 isParatext "false" @default.
- W4387074712 isRetracted "false" @default.
- W4387074712 workType "article" @default.