{"id":"https://openalex.org/W4405470172","doi":"https://doi.org/10.48550/arxiv.2412.11475","title":"OmniVLM: A Token-Compressed, Sub-Billion-Parameter Vision-Language Model for Efficient On-Device Inference","display_name":"OmniVLM: A Token-Compressed, Sub-Billion-Parameter Vision-Language Model for Efficient On-Device Inference","publication_year":2024,"publication_date":"2024-12-16","ids":{"openalex":"https://openalex.org/W4405470172","doi":"https://doi.org/10.48550/arxiv.2412.11475"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2412.11475","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2412.11475","pdf_url":"https://arxiv.org/pdf/2412.11475","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-nc-nd","license_id":"https://openalex.org/licenses/cc-by-nc-nd","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"text"},"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2412.11475","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5014261323","display_name":"Wei Chen","orcid":"https://orcid.org/0000-0002-5382-0542"},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Chen, Wei","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100337920","display_name":"Zhiyuan Li","orcid":"https://orcid.org/0000-0001-6478-1027"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Zhiyuan","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5002590940","display_name":"Shuo Xin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xin, Shuo","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":3,"corresponding_author_ids":["https://openalex.org/A5014261323"],"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/T10627","display_name":"Advanced Image and Video Retrieval Techniques","score":0.9843999743461609,"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/T10627","display_name":"Advanced Image and Video Retrieval Techniques","score":0.9843999743461609,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.983299970626831,"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/T12702","display_name":"Brain Tumor Detection and Classification","score":0.9528999924659729,"subfield":{"id":"https://openalex.org/subfields/2808","display_name":"Neurology"},"field":{"id":"https://openalex.org/fields/28","display_name":"Neuroscience"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.8214043378829956},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.7264441847801208},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.5917319655418396},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.40376636385917664},{"id":"https://openalex.org/keywords/computer-security","display_name":"Computer security","score":0.37294602394104004},{"id":"https://openalex.org/keywords/computer-vision","display_name":"Computer vision","score":0.3585541844367981},{"id":"https://openalex.org/keywords/computer-graphics","display_name":"Computer graphics (images)","score":0.3281971216201782}],"concepts":[{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.8214043378829956},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.7264441847801208},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5917319655418396},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.40376636385917664},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.37294602394104004},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.3585541844367981},{"id":"https://openalex.org/C121684516","wikidata":"https://www.wikidata.org/wiki/Q7600677","display_name":"Computer graphics (images)","level":1,"score":0.3281971216201782}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2412.11475","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2412.11475","pdf_url":"https://arxiv.org/pdf/2412.11475","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-nc-nd","license_id":"https://openalex.org/licenses/cc-by-nc-nd","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"text"},{"id":"doi:10.48550/arxiv.2412.11475","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2412.11475","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2412.11475","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2412.11475","pdf_url":"https://arxiv.org/pdf/2412.11475","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-nc-nd","license_id":"https://openalex.org/licenses/cc-by-nc-nd","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"text"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4405470172.pdf"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4388335561","https://openalex.org/W2970530566","https://openalex.org/W4288261899","https://openalex.org/W4307309205","https://openalex.org/W2967478618","https://openalex.org/W4385009901","https://openalex.org/W4385572700","https://openalex.org/W2997152889","https://openalex.org/W4285141722","https://openalex.org/W4387768015"],"abstract_inverted_index":{"We":[0],"present":[1],"OmniVLM,":[2],"a":[3,13,37,73],"sub-billion-parameter":[4],"vision-language":[5],"model":[6,108],"for":[7],"efficient":[8,102],"on-device":[9],"inference.":[10],"OmniVLM":[11,51,66],"introduces":[12],"token":[14,20],"compression":[15],"mechanism":[16],"that":[17],"reduces":[18],"visual":[19],"sequence":[21],"length":[22],"from":[23],"729":[24],"to":[25,99],"81":[26],"tokens,":[27],"significantly":[28],"reducing":[29],"computational":[30],"overhead":[31],"while":[32],"preserving":[33],"visual-semantic":[34],"fidelity.":[35],"Through":[36],"multi-stage":[38],"training":[39],"pipeline":[40],"of":[41,55],"pretraining,":[42],"supervised":[43],"fine-tuning,":[44],"and":[45,64,89,116],"minimal-edit":[46],"Direct":[47],"Preference":[48],"Optimization":[49],"(DPO),":[50],"matches":[52],"the":[53,79,117],"performance":[54],"larger":[56],"models.":[57],"On":[58],"multiple":[59],"benchmarks":[60],"including":[61],"ScienceQA,":[62],"POPE,":[63],"MMMU,":[65],"outperforms":[67],"existing":[68],"baselines":[69],"like":[70],"nanoLLAVA":[71],"within":[72],"968M-parameter":[74],"footprint.":[75],"Empirical":[76],"results":[77],"on":[78,104,113],"same":[80],"laptop":[81],"demonstrate":[82],"9.1x":[83],"faster":[84],"time-to-first-token":[85],"(0.75s":[86],"vs":[87,95],"6.82s)":[88],"1.5x":[90],"higher":[91],"decoding":[92],"speed":[93],"(29.41":[94],"19.20":[96],"tokens/s)":[97],"compared":[98],"nanoLLAVA,":[100],"enabling":[101],"deployment":[103],"edge":[105],"devices.":[106],"The":[107],"weights":[109],"can":[110,120],"be":[111,121],"accessed":[112],"huggingface:":[114],"\\url{https://huggingface.co/NexaAIDev/OmniVLM-968M},":[115],"inference":[118],"examples":[119],"find":[122],"in":[123],"Appendix":[124],"B.":[125]},"counts_by_year":[],"updated_date":"2026-03-13T16:22:10.518609","created_date":"2024-12-18T00:00:00"}