{"id":"https://openalex.org/W7133216801","doi":"https://doi.org/10.48550/arxiv.2602.23440","title":"Truncated Step-Level Sampling with Process Rewards for Retrieval-Augmented Reasoning","display_name":"Truncated Step-Level Sampling with Process Rewards for Retrieval-Augmented Reasoning","publication_year":2026,"publication_date":"2026-02-26","ids":{"openalex":"https://openalex.org/W7133216801","doi":"https://doi.org/10.48550/arxiv.2602.23440"},"language":null,"primary_location":{"id":"pmh:doi:10.48550/arxiv.2602.23440","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},"type":"article","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":null,"any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5015911813","display_name":"Chris Samarinas","orcid":"https://orcid.org/0000-0002-1941-3310"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Samarinas, Chris","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5080115221","display_name":"Haw-Shiuan Chang","orcid":"https://orcid.org/0000-0003-4607-936X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chang, Haw-Shiuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5127834456","display_name":"Hamed Zamani","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zamani, Hamed","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":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.23756541,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"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":0.560699999332428,"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":0.560699999332428,"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/T10286","display_name":"Information Retrieval and Search Behavior","score":0.10080000013113022,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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":0.09759999811649323,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/heuristic","display_name":"Heuristic","score":0.7124999761581421},{"id":"https://openalex.org/keywords/sampling","display_name":"Sampling (signal processing)","score":0.6559000015258789},{"id":"https://openalex.org/keywords/variance","display_name":"Variance (accounting)","score":0.6305999755859375},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.580299973487854},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.5626000165939331},{"id":"https://openalex.org/keywords/quality","display_name":"Quality (philosophy)","score":0.5450000166893005},{"id":"https://openalex.org/keywords/sample","display_name":"Sample (material)","score":0.48240000009536743},{"id":"https://openalex.org/keywords/importance-sampling","display_name":"Importance sampling","score":0.41780000925064087}],"concepts":[{"id":"https://openalex.org/C173801870","wikidata":"https://www.wikidata.org/wiki/Q201413","display_name":"Heuristic","level":2,"score":0.7124999761581421},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.6559000015258789},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6547999978065491},{"id":"https://openalex.org/C196083921","wikidata":"https://www.wikidata.org/wiki/Q7915758","display_name":"Variance (accounting)","level":2,"score":0.6305999755859375},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.580299973487854},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.5626000165939331},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.5450000166893005},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.539900004863739},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5224000215530396},{"id":"https://openalex.org/C198531522","wikidata":"https://www.wikidata.org/wiki/Q485146","display_name":"Sample (material)","level":2,"score":0.48240000009536743},{"id":"https://openalex.org/C52740198","wikidata":"https://www.wikidata.org/wiki/Q1539564","display_name":"Importance sampling","level":3,"score":0.41780000925064087},{"id":"https://openalex.org/C148220186","wikidata":"https://www.wikidata.org/wiki/Q7111912","display_name":"Outcome (game theory)","level":2,"score":0.3977999985218048},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.366100013256073},{"id":"https://openalex.org/C81669768","wikidata":"https://www.wikidata.org/wiki/Q2359161","display_name":"Precision and recall","level":2,"score":0.3154999911785126},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.31130000948905945},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2980000078678131},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.27619999647140503},{"id":"https://openalex.org/C2780009758","wikidata":"https://www.wikidata.org/wiki/Q6804172","display_name":"Measure (data warehouse)","level":2,"score":0.26489999890327454},{"id":"https://openalex.org/C2781039887","wikidata":"https://www.wikidata.org/wiki/Q1391724","display_name":"Factor (programming language)","level":2,"score":0.2628999948501587},{"id":"https://openalex.org/C127705205","wikidata":"https://www.wikidata.org/wiki/Q5748245","display_name":"Heuristics","level":2,"score":0.25609999895095825},{"id":"https://openalex.org/C73602740","wikidata":"https://www.wikidata.org/wiki/Q7795822","display_name":"Thompson sampling","level":3,"score":0.25450000166893005}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:doi:10.48550/arxiv.2602.23440","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},{"id":"doi:10.48550/arxiv.2602.23440","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2602.23440","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,"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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:doi:10.48550/arxiv.2602.23440","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Reinforcement":[0],"learning":[1],"has":[2],"emerged":[3],"as":[4,60],"an":[5,176],"effective":[6],"paradigm":[7],"for":[8,53,93,143,151],"training":[9],"large":[10],"language":[11],"models":[12],"to":[13,37,118,134,140],"interleave":[14],"reasoning":[15,47,164],"with":[16],"search":[17],"engine":[18],"calls.":[19],"However,":[20],"existing":[21],"approaches":[22],"face":[23],"a":[24,33,112,119,135,172,206],"fundamental":[25],"credit":[26],"assignment":[27],"problem:":[28],"methods":[29,58],"like":[30],"Search-R1":[31,211],"assign":[32],"single":[34,120],"outcome":[35,184],"reward":[36],"the":[38,81,127,146,213,219],"entire":[39],"multi-step":[40],"trajectory,":[41],"providing":[42,179],"no":[43],"signal":[44],"about":[45],"which":[46,96],"or":[48,55,186],"retrieval":[49],"decisions":[50],"were":[51],"responsible":[52],"success":[54],"failure.":[56],"Process-reward":[57],"such":[59],"StepSearch":[61],"introduce":[62],"step-level":[63,106,152,188],"supervision":[64,181],"but":[65],"still":[66],"sample":[67],"complete":[68],"trajectories":[69],"independently,":[70],"so":[71],"advantage":[72,130],"estimates":[73,131],"at":[74],"any":[75],"given":[76],"step":[77],"are":[78,223],"contaminated":[79],"by":[80,132],"randomness":[82],"of":[83,129,137],"all":[84,116],"other":[85],"steps.":[86],"We":[87,123],"propose":[88],"SLATE":[89,197],"(Step-Level":[90],"Advantage":[91],"estimation":[92],"Truncated":[94],"Exploration),":[95],"addresses":[97],"both":[98,200],"problems":[99],"through":[100],"two":[101],"complementary":[102,239],"ideas.":[103],"First,":[104],"truncated":[105,233],"sampling":[107,142,234],"generates":[108],"k":[109],"continuations":[110],"from":[111],"shared":[113],"prefix,":[114],"isolating":[115],"variation":[117],"decision":[121],"point.":[122],"prove":[124],"this":[125],"reduces":[126],"variance":[128,149],"up":[133],"factor":[136],"T":[138],"compared":[139],"full-trajectory":[141],"T-step":[144],"trajectories,":[145],"first":[147],"formal":[148],"guarantee":[150],"RL":[153],"in":[154],"retrieval-augmented":[155],"reasoning.":[156],"Second,":[157],"dense,":[158],"decomposed":[159],"process":[160],"rewards":[161,237],"separately":[162],"evaluate":[163],"quality,":[165,167],"query":[166],"and":[168,202,216,229,235],"answer":[169],"correctness":[170],"on":[171,191,212,218,225],"ternary":[173],"scale":[174],"via":[175],"LLM":[177],"judge,":[178],"richer":[180],"than":[182],"binary":[183],"signals":[185],"heuristic":[187],"scores.":[189],"Experiments":[190],"seven":[192],"QA":[193],"benchmarks":[194],"show":[195],"that":[196,232],"consistently":[198],"outperforms":[199],"sparse-reward":[201],"process-reward":[203],"baselines,":[204],"achieving":[205],"7.0%":[207],"relative":[208],"improvement":[209],"over":[210],"7B":[214],"model":[215],"30.7%":[217],"3B":[220],"model.":[221],"Gains":[222],"largest":[224],"challenging":[226],"multi-hop":[227],"tasks,":[228],"ablations":[230],"confirm":[231],"dense":[236],"provide":[238],"benefits.":[240]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-03-03T00:00:00"}