{"id":"https://openalex.org/W7114783656","doi":"https://doi.org/10.48550/arxiv.2512.08077","title":"Unveiling Latent Knowledge in Chemistry Language Models through Sparse Autoencoders","display_name":"Unveiling Latent Knowledge in Chemistry Language Models through Sparse Autoencoders","publication_year":2025,"publication_date":"2025-12-08","ids":{"openalex":"https://openalex.org/W7114783656","doi":"https://doi.org/10.48550/arxiv.2512.08077"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2512.08077","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2512.08077","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":"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.2512.08077","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Cohen, Jaron","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cohen, Jaron","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Hasson, Alexander G.","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hasson, Alexander G.","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Tanovic, Sara","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tanovic, Sara","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":0.9010000228881836,"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":0.9010000228881836,"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/T10211","display_name":"Computational Drug Discovery Methods","score":0.06729999929666519,"subfield":{"id":"https://openalex.org/subfields/1703","display_name":"Computational Theory and Mathematics"},"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":0.0044999998062849045,"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/interpretability","display_name":"Interpretability","score":0.8079000115394592},{"id":"https://openalex.org/keywords/autoencoder","display_name":"Autoencoder","score":0.5497999787330627},{"id":"https://openalex.org/keywords/foundation","display_name":"Foundation (evidence)","score":0.40709999203681946},{"id":"https://openalex.org/keywords/generative-grammar","display_name":"Generative grammar","score":0.3939000070095062},{"id":"https://openalex.org/keywords/property","display_name":"Property (philosophy)","score":0.3752000033855438},{"id":"https://openalex.org/keywords/computational-model","display_name":"Computational model","score":0.3425999879837036},{"id":"https://openalex.org/keywords/encode","display_name":"ENCODE","score":0.334199994802475},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.32359999418258667}],"concepts":[{"id":"https://openalex.org/C2781067378","wikidata":"https://www.wikidata.org/wiki/Q17027399","display_name":"Interpretability","level":2,"score":0.8079000115394592},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6259999871253967},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5627999901771545},{"id":"https://openalex.org/C101738243","wikidata":"https://www.wikidata.org/wiki/Q786435","display_name":"Autoencoder","level":3,"score":0.5497999787330627},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.484499990940094},{"id":"https://openalex.org/C2780966255","wikidata":"https://www.wikidata.org/wiki/Q5474306","display_name":"Foundation (evidence)","level":2,"score":0.40709999203681946},{"id":"https://openalex.org/C39890363","wikidata":"https://www.wikidata.org/wiki/Q36108","display_name":"Generative grammar","level":2,"score":0.3939000070095062},{"id":"https://openalex.org/C188147891","wikidata":"https://www.wikidata.org/wiki/Q147638","display_name":"Cognitive science","level":1,"score":0.38530001044273376},{"id":"https://openalex.org/C189950617","wikidata":"https://www.wikidata.org/wiki/Q937228","display_name":"Property (philosophy)","level":2,"score":0.3752000033855438},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.3580000102519989},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.35019999742507935},{"id":"https://openalex.org/C66024118","wikidata":"https://www.wikidata.org/wiki/Q1122506","display_name":"Computational model","level":2,"score":0.3425999879837036},{"id":"https://openalex.org/C66746571","wikidata":"https://www.wikidata.org/wiki/Q1134833","display_name":"ENCODE","level":3,"score":0.334199994802475},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.32359999418258667},{"id":"https://openalex.org/C68762167","wikidata":"https://www.wikidata.org/wiki/Q910164","display_name":"Cheminformatics","level":2,"score":0.3197999894618988},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.30090001225471497},{"id":"https://openalex.org/C185592680","wikidata":"https://www.wikidata.org/wiki/Q2329","display_name":"Chemistry","level":0,"score":0.2962999939918518},{"id":"https://openalex.org/C171686336","wikidata":"https://www.wikidata.org/wiki/Q3532085","display_name":"Topic model","level":2,"score":0.2953999936580658},{"id":"https://openalex.org/C179603123","wikidata":"https://www.wikidata.org/wiki/Q1941921","display_name":"Modeling language","level":3,"score":0.2944999933242798},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.27630001306533813},{"id":"https://openalex.org/C2777601683","wikidata":"https://www.wikidata.org/wiki/Q6499736","display_name":"Vocabulary","level":2,"score":0.2703999876976013},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.2648000121116638},{"id":"https://openalex.org/C74187038","wikidata":"https://www.wikidata.org/wiki/Q1418791","display_name":"Drug discovery","level":2,"score":0.2644999921321869},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.25929999351501465},{"id":"https://openalex.org/C161301231","wikidata":"https://www.wikidata.org/wiki/Q3478658","display_name":"Knowledge representation and reasoning","level":2,"score":0.257999986410141},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.25380000472068787}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2512.08077","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2512.08077","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":"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.2512.08077","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2512.08077","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":"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/4","display_name":"Quality Education","score":0.7405461668968201}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Since":[0],"the":[1,81,167],"advent":[2],"of":[3,116,129],"machine":[4],"learning,":[5],"interpretability":[6],"has":[7,157],"remained":[8],"a":[9,113,144],"persistent":[10],"challenge,":[11],"becoming":[12],"increasingly":[13],"urgent":[14],"as":[15],"generative":[16],"models":[17,38,53,111],"support":[18],"high-stakes":[19],"applications":[20],"in":[21,28,43,151],"drug":[22,139],"and":[23,47,71,97,126,137,163],"material":[24],"discovery.":[25],"Recent":[26],"advances":[27],"large":[29],"language":[30,37],"model":[31],"(LLM)":[32],"architectures":[33],"have":[34],"yielded":[35],"chemistry":[36,88,172],"(CLMs)":[39],"with":[40,166],"impressive":[41],"capabilities":[42],"molecular":[44,48,104],"property":[45],"prediction":[46],"generation.":[49],"However,":[50],"how":[51],"these":[52,110],"internally":[54],"represent":[55],"chemical":[56,117,130],"knowledge":[57,150],"remains":[58],"poorly":[59],"understood.":[60],"In":[61],"this":[62],"work,":[63],"we":[64,91],"extend":[65],"sparse":[66],"autoencoder":[67],"techniques":[68],"to":[69,80,169],"uncover":[70],"examine":[72],"interpretable":[73],"features":[74,96,125],"within":[75],"CLMs.":[76],"Applying":[77],"our":[78],"methodology":[79],"Foundation":[82],"Models":[83],"for":[84,147,159],"Materials":[85],"(FM4M)":[86],"SMI-TED":[87],"foundation":[89],"model,":[90],"extract":[92],"semantically":[93],"meaningful":[94],"latent":[95,124,149],"analyse":[98],"their":[99],"activation":[100],"patterns":[101],"across":[102],"diverse":[103],"datasets.":[105],"Our":[106,141],"findings":[107],"reveal":[108],"that":[109],"encode":[112],"rich":[114],"landscape":[115],"concepts.":[118],"We":[119],"identify":[120],"correlations":[121],"between":[122],"specific":[123],"distinct":[127],"domains":[128],"knowledge,":[131],"including":[132],"structural":[133],"motifs,":[134],"physicochemical":[135],"properties,":[136],"pharmacological":[138],"classes.":[140],"approach":[142],"provides":[143],"generalisable":[145],"framework":[146],"uncovering":[148],"chemistry-focused":[152],"AI":[153],"systems.":[154],"This":[155],"work":[156],"implications":[158],"both":[160],"foundational":[161],"understanding":[162],"practical":[164],"deployment;":[165],"potential":[168],"accelerate":[170],"computational":[171],"research.":[173]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2025-12-11T00:00:00"}