{"id":"https://openalex.org/W7160913662","doi":"https://doi.org/10.48550/arxiv.2605.10556","title":"EnergyLens: Interpretable Closed-Form Energy Models for Multimodal LLM Inference Serving","display_name":"EnergyLens: Interpretable Closed-Form Energy Models for Multimodal LLM Inference Serving","publication_year":2026,"publication_date":"2026-05-11","ids":{"openalex":"https://openalex.org/W7160913662","doi":"https://doi.org/10.48550/arxiv.2605.10556"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.10556","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.10556","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.2605.10556","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5104492967","display_name":"Vittorio Palladino","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Palladino, Vittorio","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135953491","display_name":"Gianluca Palermo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Palermo, Gianluca","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135960556","display_name":"Michael E. Papka","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Papka, Michael E.","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5135944016","display_name":"Zhiling Lan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lan, Zhiling","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/T11948","display_name":"Machine Learning in Materials Science","score":0.10769999772310257,"subfield":{"id":"https://openalex.org/subfields/2505","display_name":"Materials Chemistry"},"field":{"id":"https://openalex.org/fields/25","display_name":"Materials Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11948","display_name":"Machine Learning in Materials Science","score":0.10769999772310257,"subfield":{"id":"https://openalex.org/subfields/2505","display_name":"Materials Chemistry"},"field":{"id":"https://openalex.org/fields/25","display_name":"Materials Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.08959999680519104,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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":0.08799999952316284,"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/profiling","display_name":"Profiling (computer programming)","score":0.7286999821662903},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6708999872207642},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.4352000057697296},{"id":"https://openalex.org/keywords/efficient-energy-use","display_name":"Efficient energy use","score":0.3635999858379364},{"id":"https://openalex.org/keywords/model-selection","display_name":"Model selection","score":0.34130001068115234},{"id":"https://openalex.org/keywords/proxy","display_name":"Proxy (statistics)","score":0.3249000012874603},{"id":"https://openalex.org/keywords/pipeline","display_name":"Pipeline (software)","score":0.30630001425743103}],"concepts":[{"id":"https://openalex.org/C187191949","wikidata":"https://www.wikidata.org/wiki/Q1138496","display_name":"Profiling (computer programming)","level":2,"score":0.7286999821662903},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7113000154495239},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6708999872207642},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.44780001044273376},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.4352000057697296},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.41769999265670776},{"id":"https://openalex.org/C2742236","wikidata":"https://www.wikidata.org/wiki/Q924713","display_name":"Efficient energy use","level":2,"score":0.3635999858379364},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.35920000076293945},{"id":"https://openalex.org/C93959086","wikidata":"https://www.wikidata.org/wiki/Q6888345","display_name":"Model selection","level":2,"score":0.34130001068115234},{"id":"https://openalex.org/C2780148112","wikidata":"https://www.wikidata.org/wiki/Q1432581","display_name":"Proxy (statistics)","level":2,"score":0.3249000012874603},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.30630001425743103},{"id":"https://openalex.org/C83546350","wikidata":"https://www.wikidata.org/wiki/Q1139051","display_name":"Regression","level":2,"score":0.2928999960422516},{"id":"https://openalex.org/C2779714256","wikidata":"https://www.wikidata.org/wiki/Q25305062","display_name":"Multiple Models","level":2,"score":0.2791999876499176},{"id":"https://openalex.org/C3019966295","wikidata":"https://www.wikidata.org/wiki/Q1341368","display_name":"Energy cost","level":2,"score":0.2775999903678894},{"id":"https://openalex.org/C186370098","wikidata":"https://www.wikidata.org/wiki/Q442787","display_name":"Energy (signal processing)","level":2,"score":0.27309998869895935},{"id":"https://openalex.org/C45804977","wikidata":"https://www.wikidata.org/wiki/Q7239673","display_name":"Predictive modelling","level":2,"score":0.26489999890327454},{"id":"https://openalex.org/C160920958","wikidata":"https://www.wikidata.org/wiki/Q7662746","display_name":"Synthetic data","level":2,"score":0.251800000667572},{"id":"https://openalex.org/C2989134064","wikidata":"https://www.wikidata.org/wiki/Q288510","display_name":"Execution time","level":2,"score":0.2502000033855438}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.10556","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.10556","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.2605.10556","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.10556","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":[{"id":"https://metadata.un.org/sdg/7","display_name":"Affordable and clean energy","score":0.8480322957038879}],"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],"large":[1],"language":[2],"models":[3],"span":[4],"dense,":[5],"mixture-of-experts,":[6],"and":[7,10,31,55,66,79,116,125,129,140,183,190],"state-space":[8],"architectures":[9],"are":[11],"deployed":[12],"on":[13,44],"heterogeneous":[14],"accelerators":[15],"under":[16,50],"increasingly":[17],"diverse":[18],"multimodal":[19],"workloads,":[20],"optimising":[21],"inference":[22],"energy":[23,40,56,101],"has":[24],"become":[25],"as":[26,28,38,88,110,144,146],"critical":[27],"optimizing":[29],"latency":[30,37,54],"throughput.":[32],"Existing":[33],"approaches":[34],"either":[35],"treat":[36],"an":[39],"proxy":[41],"or":[42],"rely":[43],"data-hungry":[45],"black-box":[46,67,120],"surrogates.":[47],"Both":[48],"fail":[49],"varying":[51],"parallelism":[52,127],"strategies:":[53],"optima":[57],"diverge":[58],"in":[59,104],"over":[60,92],"20%":[61],"of":[62,71,106,112,174],"configurations":[63],"we":[64],"tested,":[65],"surrogates":[68],"require":[69],"hundreds":[70],"profiling":[72,93,148,181],"samples":[73],"to":[74,95,162,186],"generalize":[75],"across":[76,157],"model":[77,102],"families":[78],"hardware.":[80],"We":[81],"present":[82],"EnergyLens,":[83],"which":[84],"uses":[85],"symbolic":[86],"regression":[87],"a":[89,97,198],"structure-discovery":[90],"tool":[91,201],"data":[94],"derive":[96],"single":[98],"twelve-parameter":[99],"closed-form":[100],"expressed":[103],"terms":[105],"system":[107],"properties":[108],"such":[109],"degree":[111],"parallelism,":[113],"batch":[114,188],"size,":[115],"sequence":[117],"length.":[118],"Unlike":[119],"surrogates,":[121],"EnergyLens":[122,150],"decouples":[123],"tensor":[124],"pipeline":[126],"contributions":[128],"separates":[130],"prefill":[131],"from":[132,143],"decode":[133],"energy,":[134],"making":[135,196],"its":[136],"predictions":[137],"physically":[138],"interpretable":[139,200],"actionable.":[141],"Fitted":[142],"few":[145],"50":[147],"measurements,":[149],"achieves":[151],"88.2%":[152],"Top-1":[153],"configuration":[154],"selection":[155],"accuracy":[156,173],"many":[158],"evaluation":[159],"scenarios":[160],"compared":[161],"60.9%":[163],"for":[164,202],"the":[165,171],"closest":[166],"prior":[167],"analytical":[168],"baseline,":[169],"matches":[170],"predictive":[172],"ensemble":[175],"ML":[176],"methods":[177],"with":[178],"10x":[179],"fewer":[180],"samples,":[182],"extrapolates":[184],"reliably":[185],"unseen":[187],"sizes":[189],"hardware":[191],"platforms":[192],"without":[193],"structural":[194],"modification,":[195],"it":[197],"practical,":[199],"energy-optimal":[203],"LLM":[204],"deployment.":[205]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-13T00:00:00"}