{"id":"https://openalex.org/W4405033595","doi":"https://doi.org/10.48550/arxiv.2412.00821","title":"Improving Physics Reasoning in Large Language Models Using Mixture of Refinement Agents","display_name":"Improving Physics Reasoning in Large Language Models Using Mixture of Refinement Agents","publication_year":2024,"publication_date":"2024-12-01","ids":{"openalex":"https://openalex.org/W4405033595","doi":"https://doi.org/10.48550/arxiv.2412.00821"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2412.00821","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2412.00821","pdf_url":"https://arxiv.org/pdf/2412.00821","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2412.00821","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5078468599","display_name":"Raj Jaiswal","orcid":"https://orcid.org/0009-0008-5044-1409","observed_orcids":["https://orcid.org/0009-0008-5044-1409"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jaiswal, Raj","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Jain, Dhruv","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jain, Dhruv","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Popat, Harsh Parimal","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Popat, Harsh Parimal","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5010012157","display_name":"Avinash Anand","orcid":"https://orcid.org/0009-0003-2479-0342","observed_orcids":["https://orcid.org/0009-0003-2479-0342"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Anand, Avinash","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5093034547","display_name":"Abhishek Dharmadhikari","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dharmadhikari, Abhishek","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113036777","display_name":"Atharva Marathe","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Marathe, Atharva","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5079357056","display_name":"Rajiv Ratn Shah","orcid":"https://orcid.org/0000-0003-1028-9373","observed_orcids":["https://orcid.org/0000-0003-1028-9373"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shah, Rajiv Ratn","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.5601000189781189,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.5601000189781189,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.14149999618530273,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.04919999837875366,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/physics-reasoning","display_name":"physics reasoning","score":0.8100000023841858},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.9559999704360962},{"id":"https://openalex.org/keywords/scientific-reasoning","display_name":"scientific reasoning","score":0.6880000233650208},{"id":"https://openalex.org/keywords/computational-errors","display_name":"computational errors","score":0.9419999718666077},{"id":"https://openalex.org/keywords/open-source-llms","display_name":"open-source LLMs","score":0.8159999847412109},{"id":"https://openalex.org/keywords/llama-3-70b","display_name":"Llama-3-70B","score":0.8709999918937683},{"id":"https://openalex.org/keywords/gpt-4o","display_name":"GPT-4o","score":0.8199999928474426}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4466349482536316}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2412.00821","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2412.00821","pdf_url":"https://arxiv.org/pdf/2412.00821","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2412.00821","doi":"https://doi.org/10.48550/arxiv.2412.00821","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2412.00821","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2412.00821","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2412.00821","pdf_url":"https://arxiv.org/pdf/2412.00821","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4405033595.pdf","grobid_xml":"https://content.openalex.org/works/W4405033595.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4405626639","https://openalex.org/W4405254614","https://openalex.org/W7166883808","https://openalex.org/W4416548186","https://openalex.org/W4410280157","https://openalex.org/W4416364017","https://openalex.org/W4416059733","https://openalex.org/W7161715468","https://openalex.org/W4414456955","https://openalex.org/W4414858746"],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"demonstrate":[4],"remarkable":[5],"capabilities":[6],"in":[7,23,109,176],"various":[8],"reasoning":[9,30],"tasks.":[10],"However,":[11],"they":[12],"encounter":[13],"significant":[14,111],"challenges":[15],"when":[16],"it":[17],"comes":[18],"to":[19,120,136,172],"scientific":[20],"reasoning,":[21],"particularly":[22],"physics,":[24],"which":[25],"requires":[26],"not":[27],"only":[28],"mathematical":[29],"but":[31],"also":[32],"factual":[33],"and":[34,53,127,148,165],"conceptual":[35],"understanding.":[36],"When":[37],"addressing":[38],"complex":[39],"physics":[40,155],"problems,":[41],"LLMs":[42,126],"typically":[43],"face":[44],"three":[45,77],"key":[46],"issues:":[47],"problem":[48],"miscomprehension,":[49],"incorrect":[50],"concept":[51],"application,":[52],"computational":[54],"errors.":[55],"While":[56],"each":[57],"of":[58,86,163],"these":[59,138,168],"problems":[60],"can":[61,74],"be":[62],"addressed":[63],"individually,":[64],"there":[65],"is":[66],"a":[67,70,90,110,173],"need":[68],"for":[69,114],"generalized":[71],"approach":[72,118,144],"that":[73,95],"tackle":[75],"all":[76],"issues":[78],"simultaneously.":[79],"To":[80],"address":[81],"this,":[82],"we":[83],"introduce":[84],"Mixture":[85],"Refinement":[87],"Agents":[88],"(MoRA),":[89],"novel":[91],"agentic":[92],"refinement":[93,139],"framework":[94],"iteratively":[96],"refines":[97],"the":[98,105,122,131,146,161],"LLM":[99],"generated":[100],"base":[101],"solution":[102],"by":[103,129],"correcting":[104],"aforementioned":[106],"errors,":[107],"resulting":[108],"performance":[112,162],"improvement":[113],"open-source":[115],"LLMs.":[116],"Our":[117],"aims":[119],"bridge":[121],"gap":[123],"between":[124],"opensource":[125],"GPT-4o":[128],"utilizing":[130],"latter":[132],"as":[133],"error":[134],"identifier":[135],"guide":[137],"agents.":[140],"We":[141],"evaluate":[142],"our":[143,153],"on":[145,167],"SciEval":[147],"MMLU":[149],"subsets":[150],"along":[151],"with":[152],"own":[154],"dataset":[156],"(PhysicsQA).":[157],"MoRA":[158],"significantly":[159],"improves":[160],"Llama-3-70B":[164],"Gemma-2-27B":[166],"datasets,":[169],"achieving":[170],"up":[171],"16%":[174],"increase":[175],"final":[177],"answer":[178],"accuracy.":[179]},"counts_by_year":[],"updated_date":"2026-10-01T08:54:57.152042","created_date":"2024-12-05T00:00:00"}