{"id":"https://openalex.org/W7162684734","doi":"https://doi.org/10.48550/arxiv.2605.28556","title":"A Matter of TASTE: Improving Coverage and Difficulty of Agent Benchmarks","display_name":"A Matter of TASTE: Improving Coverage and Difficulty of Agent Benchmarks","publication_year":2026,"publication_date":"2026-05-27","ids":{"openalex":"https://openalex.org/W7162684734","doi":"https://doi.org/10.48550/arxiv.2605.28556"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.28556","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.28556","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"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"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2605.28556","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5057783696","display_name":"Tomer Keren","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Keren, Tomer","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137292898","display_name":"Nitay Calderon","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Calderon, Nitay","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5024163531","display_name":"Asaf Yehudai","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yehudai, Asaf","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137219800","display_name":"Yotam Perlitz","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Perlitz, Yotam","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5051536706","display_name":"Michal Shmueli-Scheuer","orcid":"https://orcid.org/0000-0002-6386-8726"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shmueli-Scheuer, Michal","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5137284446","display_name":"Roi Reichert","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Reichart, Roi","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":false,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.12520000338554382,"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"}},"topics":[{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.12520000338554382,"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/T10028","display_name":"Topic Modeling","score":0.0794999971985817,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.06530000269412994,"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/benchmark","display_name":"Benchmark (surveying)","score":0.646399974822998},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.5856000185012817},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.5620999932289124},{"id":"https://openalex.org/keywords/construct","display_name":"Construct (python library)","score":0.48339998722076416},{"id":"https://openalex.org/keywords/cover","display_name":"Cover (algebra)","score":0.3653999865055084},{"id":"https://openalex.org/keywords/range","display_name":"Range (aeronautics)","score":0.35440000891685486}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7860000133514404},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.646399974822998},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.5856000185012817},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.5620999932289124},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4925999939441681},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.48559999465942383},{"id":"https://openalex.org/C2780801425","wikidata":"https://www.wikidata.org/wiki/Q5164392","display_name":"Construct (python library)","level":2,"score":0.48339998722076416},{"id":"https://openalex.org/C2780428219","wikidata":"https://www.wikidata.org/wiki/Q16952335","display_name":"Cover (algebra)","level":2,"score":0.3653999865055084},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.35440000891685486},{"id":"https://openalex.org/C2778112365","wikidata":"https://www.wikidata.org/wiki/Q3511065","display_name":"Sequence (biology)","level":2,"score":0.32710000872612},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.3228999972343445},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.2913999855518341},{"id":"https://openalex.org/C86251818","wikidata":"https://www.wikidata.org/wiki/Q816754","display_name":"Benchmarking","level":2,"score":0.2711000144481659},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.26930001378059387},{"id":"https://openalex.org/C112972136","wikidata":"https://www.wikidata.org/wiki/Q7595718","display_name":"Stability (learning theory)","level":2,"score":0.25690001249313354}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.28556","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.28556","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.28556","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.28556","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"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"},"sustainable_development_goals":[{"display_name":"Decent work and economic growth","score":0.6745250225067139,"id":"https://metadata.un.org/sdg/8"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"As":[0],"agent":[1],"capabilities":[2],"advance,":[3],"existing":[4,200],"benchmarks,":[5,217],"such":[6],"as":[7],"$\u03c4^2$-Bench,":[8],"are":[9,30],"becoming":[10],"increasingly":[11],"saturated.":[12],"Yet":[13],"constructing":[14],"new":[15],"benchmark":[16,126],"tasks":[17,81,168,181],"remains":[18],"complex,":[19],"costly,":[20],"and":[21,36,128,155],"labor-intensive.":[22],"Moreover,":[23],"the":[24,48,62,118,144,185,212],"standard":[25],"approach,":[26],"in":[27,33],"which":[28],"scenarios":[29],"first":[31],"written":[32],"natural":[34],"language":[35],"then":[37,113],"mapped":[38],"to":[39,174],"tool":[40,102,110,189],"sequences,":[41],"captures":[42],"only":[43],"a":[44,106,140],"narrow":[45],"subset":[46],"of":[47,109,143,147,187,214,223],"tool-use":[49,84],"patterns":[50],"agents":[51,191],"exercise.":[52],"In":[53],"this":[54],"paper,":[55],"we":[56,137],"address":[57],"these":[58],"problems":[59],"by":[60],"reversing":[61],"task":[63],"construction":[64],"process.":[65],"We":[66,149],"propose":[67],"TASTE:":[68],"Task":[69],"Synthesis":[70],"from":[71,117,172],"Tool":[72],"Sequence":[73],"Evolution,":[74],"an":[75,88],"automatic":[76],"method":[77],"that":[78,104,157],"generates":[79],"challenging":[80,141],"with":[82],"broader":[83],"coverage.":[85],"TASTE":[86,112,218],"utilizes":[87],"Adaptive":[89],"Contrastive":[90],"$n$-gram":[91],"model":[92],"trained":[93],"on":[94,166,199],"LLM-judged":[95],"validity":[96],"signals.":[97],"This":[98],"enables":[99,219],"sampling":[100],"valid":[101],"sequences":[103,116],"cover":[105],"vast":[107],"range":[108],"combinations.":[111],"selects":[114],"representative":[115],"pool":[119],"via":[120],"clustering,":[121],"instantiates":[122],"them":[123,130],"into":[124],"complete":[125],"tasks,":[127],"refines":[129],"through":[131],"iterative":[132],"difficulty":[133],"evolution.":[134],"Using":[135],"TASTE,":[136],"construct":[138],"$\u03c4^c$-Bench,":[139],"extension":[142],"three":[145],"domains":[146],"$\u03c4^2$-Bench.":[148],"evaluate":[150],"$11$":[151],"agent/user":[152],"LLM":[153],"pairs":[154],"find":[156],"models":[158],"nearly":[159],"saturating":[160],"$\u03c4^2$-Bench":[161],"suffer":[162],"severe":[163],"performance":[164],"drops":[165],"our":[167,179],"(e.g.,":[169],"Gemini-3-Flash":[170],"falls":[171],"$0.82\\!-\\!0.94$":[173],"$0.28\\!-\\!0.61$).":[175],"Beyond":[176],"increasing":[177],"difficulty,":[178],"generated":[180],"more":[182],"than":[183,206],"double":[184],"number":[186],"unique":[188],"combinations":[190],"must":[192],"execute.":[193],"Our":[194],"results":[195],"suggest":[196],"high":[197],"scores":[198],"benchmarks":[201],"often":[202],"reflect":[203],"saturation":[204],"rather":[205],"robust":[207],"task-solving":[208],"ability.":[209],"By":[210],"automating":[211],"generation":[213],"difficult,":[215],"high-coverage":[216],"continuous,":[220],"scalable":[221],"evaluation":[222],"future":[224],"agents.":[225]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-29T00:00:00"}