Matches in SemOpenAlex for { <https://semopenalex.org/work/W4387560111> ?p ?o ?g. }
Showing items 1 to 69 of
69
with 100 items per page.
- W4387560111 abstract "In math reasoning with large language models (LLMs), fine-tuning data augmentation by query evolution and diverse reasoning paths is empirically verified effective, profoundly narrowing the gap between open-sourced LLMs and cutting-edge proprietary LLMs. In this paper, we conduct an investigation for such data augmentation in math reasoning and are intended to answer: (1) What strategies of data augmentation are more effective; (2) What is the scaling relationship between the amount of augmented data and model performance; and (3) Can data augmentation incentivize generalization to out-of-domain mathematical reasoning tasks? To this end, we create a new dataset, AugGSM8K, by complicating and diversifying the queries from GSM8K and sampling multiple reasoning paths. We obtained a series of LLMs called MuggleMath by fine-tuning on subsets of AugGSM8K. MuggleMath substantially achieves new state-of-the-art on GSM8K (from 54% to 68.4% at the scale of 7B, and from 63.9% to 74.0% at the scale of 13B). A log-linear relationship is presented between MuggleMath's performance and the amount of augmented data. We also find that MuggleMath is weak in out-of-domain math reasoning generalization to MATH. This is attributed to the differences in query distribution between AugGSM8K and MATH which suggest that augmentation on a single benchmark could not help with overall math reasoning performance. Codes and AugGSM8K will be uploaded to https://github.com/OFA-Sys/gsm8k-ScRel." @default.
- W4387560111 created "2023-10-12" @default.
- W4387560111 creator A5006787626 @default.
- W4387560111 creator A5008320336 @default.
- W4387560111 creator A5039820086 @default.
- W4387560111 creator A5041916202 @default.
- W4387560111 creator A5056808064 @default.
- W4387560111 creator A5065907658 @default.
- W4387560111 creator A5075695432 @default.
- W4387560111 creator A5091103295 @default.
- W4387560111 date "2023-10-09" @default.
- W4387560111 modified "2023-10-18" @default.
- W4387560111 title "Query and Response Augmentation Cannot Help Out-of-domain Math Reasoning Generalization" @default.
- W4387560111 doi "https://doi.org/10.48550/arxiv.2310.05506" @default.
- W4387560111 hasPublicationYear "2023" @default.
- W4387560111 type Work @default.
- W4387560111 citedByCount "0" @default.
- W4387560111 crossrefType "posted-content" @default.
- W4387560111 hasAuthorship W4387560111A5006787626 @default.
- W4387560111 hasAuthorship W4387560111A5008320336 @default.
- W4387560111 hasAuthorship W4387560111A5039820086 @default.
- W4387560111 hasAuthorship W4387560111A5041916202 @default.
- W4387560111 hasAuthorship W4387560111A5056808064 @default.
- W4387560111 hasAuthorship W4387560111A5065907658 @default.
- W4387560111 hasAuthorship W4387560111A5075695432 @default.
- W4387560111 hasAuthorship W4387560111A5091103295 @default.
- W4387560111 hasBestOaLocation W43875601111 @default.
- W4387560111 hasConcept C111919701 @default.
- W4387560111 hasConcept C121332964 @default.
- W4387560111 hasConcept C13280743 @default.
- W4387560111 hasConcept C134306372 @default.
- W4387560111 hasConcept C177148314 @default.
- W4387560111 hasConcept C185798385 @default.
- W4387560111 hasConcept C205649164 @default.
- W4387560111 hasConcept C2778755073 @default.
- W4387560111 hasConcept C33923547 @default.
- W4387560111 hasConcept C36503486 @default.
- W4387560111 hasConcept C41008148 @default.
- W4387560111 hasConcept C62520636 @default.
- W4387560111 hasConcept C71901391 @default.
- W4387560111 hasConceptScore W4387560111C111919701 @default.
- W4387560111 hasConceptScore W4387560111C121332964 @default.
- W4387560111 hasConceptScore W4387560111C13280743 @default.
- W4387560111 hasConceptScore W4387560111C134306372 @default.
- W4387560111 hasConceptScore W4387560111C177148314 @default.
- W4387560111 hasConceptScore W4387560111C185798385 @default.
- W4387560111 hasConceptScore W4387560111C205649164 @default.
- W4387560111 hasConceptScore W4387560111C2778755073 @default.
- W4387560111 hasConceptScore W4387560111C33923547 @default.
- W4387560111 hasConceptScore W4387560111C36503486 @default.
- W4387560111 hasConceptScore W4387560111C41008148 @default.
- W4387560111 hasConceptScore W4387560111C62520636 @default.
- W4387560111 hasConceptScore W4387560111C71901391 @default.
- W4387560111 hasLocation W43875601111 @default.
- W4387560111 hasOpenAccess W4387560111 @default.
- W4387560111 hasPrimaryLocation W43875601111 @default.
- W4387560111 hasRelatedWork W2003209439 @default.
- W4387560111 hasRelatedWork W2358319515 @default.
- W4387560111 hasRelatedWork W2378211422 @default.
- W4387560111 hasRelatedWork W2497626292 @default.
- W4387560111 hasRelatedWork W2967006609 @default.
- W4387560111 hasRelatedWork W2972592048 @default.
- W4387560111 hasRelatedWork W3037018281 @default.
- W4387560111 hasRelatedWork W4321353415 @default.
- W4387560111 hasRelatedWork W4321854979 @default.
- W4387560111 hasRelatedWork W2944823289 @default.
- W4387560111 isParatext "false" @default.
- W4387560111 isRetracted "false" @default.
- W4387560111 workType "article" @default.