{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:X2JH3KM74GI2I26CDQLZH2OVY3","short_pith_number":"pith:X2JH3KM7","schema_version":"1.0","canonical_sha256":"be927da99fe191a46bc21c1793e9d5c6ef1ff0b39ec941cdb7b6f9ad02d4acff","source":{"kind":"arxiv","id":"2506.06884","version":1},"attestation_state":"computed","paper":{"title":"FREE: Fast and Robust Vision Language Models with Early Exits","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Divya Jyoti Bajpai, Manjesh Kumar Hanawal","submitted_at":"2025-06-07T18:26:58Z","abstract_excerpt":"In recent years, Vision-Language Models (VLMs) have shown remarkable performance improvements in Vision-Language tasks. However, their large size poses challenges for real-world applications where inference latency is a concern. To tackle this issue, we propose employing Early Exit (EE) strategies in VLMs. However, training exit classifiers in VLMs is challenging, particularly with limited labeled training data. To address this, we introduce FREE, an adversarial training approach within a GAN-based framework. Here, each exit consists of a transformer layer and a classifier. The transformer lay"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2506.06884","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-06-07T18:26:58Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"c74ec85cfe11023459a2089ce5057aab952e69b5928ffa2ebcd6ea2e6a2db626","abstract_canon_sha256":"d6d7f7e38a565dbc1c8c5b287b45481e098ce68dd7aa680548934e5fd2308c21"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:17:57.508008Z","signature_b64":"VXz8J0/BMm1pcBxZfQXwspY2FMTiuT1+riRhkDrKbPtx8aj9zxZYbFMsz0eP3UpKv1WcuWBbjU35aGtiDBPBCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"be927da99fe191a46bc21c1793e9d5c6ef1ff0b39ec941cdb7b6f9ad02d4acff","last_reissued_at":"2026-07-05T11:17:57.507241Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:17:57.507241Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"FREE: Fast and Robust Vision Language Models with Early Exits","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Divya Jyoti Bajpai, Manjesh Kumar Hanawal","submitted_at":"2025-06-07T18:26:58Z","abstract_excerpt":"In recent years, Vision-Language Models (VLMs) have shown remarkable performance improvements in Vision-Language tasks. However, their large size poses challenges for real-world applications where inference latency is a concern. To tackle this issue, we propose employing Early Exit (EE) strategies in VLMs. However, training exit classifiers in VLMs is challenging, particularly with limited labeled training data. To address this, we introduce FREE, an adversarial training approach within a GAN-based framework. Here, each exit consists of a transformer layer and a classifier. The transformer lay"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.06884","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2506.06884/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2506.06884","created_at":"2026-07-05T11:17:57.507335+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.06884v1","created_at":"2026-07-05T11:17:57.507335+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.06884","created_at":"2026-07-05T11:17:57.507335+00:00"},{"alias_kind":"pith_short_12","alias_value":"X2JH3KM74GI2","created_at":"2026-07-05T11:17:57.507335+00:00"},{"alias_kind":"pith_short_16","alias_value":"X2JH3KM74GI2I26C","created_at":"2026-07-05T11:17:57.507335+00:00"},{"alias_kind":"pith_short_8","alias_value":"X2JH3KM7","created_at":"2026-07-05T11:17:57.507335+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/X2JH3KM74GI2I26CDQLZH2OVY3","json":"https://pith.science/pith/X2JH3KM74GI2I26CDQLZH2OVY3.json","graph_json":"https://pith.science/api/pith-number/X2JH3KM74GI2I26CDQLZH2OVY3/graph.json","events_json":"https://pith.science/api/pith-number/X2JH3KM74GI2I26CDQLZH2OVY3/events.json","paper":"https://pith.science/paper/X2JH3KM7"},"agent_actions":{"view_html":"https://pith.science/pith/X2JH3KM74GI2I26CDQLZH2OVY3","download_json":"https://pith.science/pith/X2JH3KM74GI2I26CDQLZH2OVY3.json","view_paper":"https://pith.science/paper/X2JH3KM7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.06884&json=true","fetch_graph":"https://pith.science/api/pith-number/X2JH3KM74GI2I26CDQLZH2OVY3/graph.json","fetch_events":"https://pith.science/api/pith-number/X2JH3KM74GI2I26CDQLZH2OVY3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/X2JH3KM74GI2I26CDQLZH2OVY3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/X2JH3KM74GI2I26CDQLZH2OVY3/action/storage_attestation","attest_author":"https://pith.science/pith/X2JH3KM74GI2I26CDQLZH2OVY3/action/author_attestation","sign_citation":"https://pith.science/pith/X2JH3KM74GI2I26CDQLZH2OVY3/action/citation_signature","submit_replication":"https://pith.science/pith/X2JH3KM74GI2I26CDQLZH2OVY3/action/replication_record"}},"created_at":"2026-07-05T11:17:57.507335+00:00","updated_at":"2026-07-05T11:17:57.507335+00:00"}