{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:N3FM7D3NI5S4XDLFOWLG3YIU43","short_pith_number":"pith:N3FM7D3N","schema_version":"1.0","canonical_sha256":"6ecacf8f6d4765cb8d6575966de114e6f292ec3fa63660acae1ed39d405af8ad","source":{"kind":"arxiv","id":"2502.02069","version":1},"attestation_state":"computed","paper":{"title":"LoRA-TTT: Low-Rank Test-Time Training for Vision-Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Jiarui Xu, Xiaolong Wang, Xueyan Zou, Yuto Kojima","submitted_at":"2025-02-04T07:40:26Z","abstract_excerpt":"The rapid advancements in vision-language models (VLMs), such as CLIP, have intensified the need to address distribution shifts between training and testing datasets. Although prior Test-Time Training (TTT) techniques for VLMs have demonstrated robust performance, they predominantly rely on tuning text prompts, a process that demands substantial computational resources and is heavily dependent on entropy-based loss. In this paper, we propose LoRA-TTT, a novel TTT method that leverages Low-Rank Adaptation (LoRA), applied exclusively to the image encoder of VLMs. By introducing LoRA and updating"},"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":"2502.02069","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-02-04T07:40:26Z","cross_cats_sorted":[],"title_canon_sha256":"9697daecbe65779d120daebdfe71fb6c6946c32fb84df88dabd67cb33d619308","abstract_canon_sha256":"118338decdffa06f8793d4b8d6f31cde5d9bf4bca2e7953679be11b4c60287ba"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:09:23.290351Z","signature_b64":"HbbgIHHq00vNNOzQMiQ6JDHaOkm4T8iIhHRL5sxyJLFHbcgk6p7tXo/Njk6A0RG/KY+DurxE4CXsPSkjvbHTCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6ecacf8f6d4765cb8d6575966de114e6f292ec3fa63660acae1ed39d405af8ad","last_reissued_at":"2026-07-05T10:09:23.289871Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:09:23.289871Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LoRA-TTT: Low-Rank Test-Time Training for Vision-Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Jiarui Xu, Xiaolong Wang, Xueyan Zou, Yuto Kojima","submitted_at":"2025-02-04T07:40:26Z","abstract_excerpt":"The rapid advancements in vision-language models (VLMs), such as CLIP, have intensified the need to address distribution shifts between training and testing datasets. Although prior Test-Time Training (TTT) techniques for VLMs have demonstrated robust performance, they predominantly rely on tuning text prompts, a process that demands substantial computational resources and is heavily dependent on entropy-based loss. In this paper, we propose LoRA-TTT, a novel TTT method that leverages Low-Rank Adaptation (LoRA), applied exclusively to the image encoder of VLMs. By introducing LoRA and updating"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.02069","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/2502.02069/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":"2502.02069","created_at":"2026-07-05T10:09:23.289928+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.02069v1","created_at":"2026-07-05T10:09:23.289928+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.02069","created_at":"2026-07-05T10:09:23.289928+00:00"},{"alias_kind":"pith_short_12","alias_value":"N3FM7D3NI5S4","created_at":"2026-07-05T10:09:23.289928+00:00"},{"alias_kind":"pith_short_16","alias_value":"N3FM7D3NI5S4XDLF","created_at":"2026-07-05T10:09:23.289928+00:00"},{"alias_kind":"pith_short_8","alias_value":"N3FM7D3N","created_at":"2026-07-05T10:09:23.289928+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.26457","citing_title":"A probabilistic framework for online test-time adaptation","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2606.23521","citing_title":"Concordia: JIT-Compiled Persistent-Kernel Checkpointing for Fault-Tolerant LLM Inference","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28142","citing_title":"MixTTA: Low-Rank Cross-Channel Mixing for Reliable Test-Time Adaptation","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28142","citing_title":"MixTTA: Low-Rank Cross-Channel Mixing for Reliable Test-Time Adaptation","ref_index":42,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/N3FM7D3NI5S4XDLFOWLG3YIU43","json":"https://pith.science/pith/N3FM7D3NI5S4XDLFOWLG3YIU43.json","graph_json":"https://pith.science/api/pith-number/N3FM7D3NI5S4XDLFOWLG3YIU43/graph.json","events_json":"https://pith.science/api/pith-number/N3FM7D3NI5S4XDLFOWLG3YIU43/events.json","paper":"https://pith.science/paper/N3FM7D3N"},"agent_actions":{"view_html":"https://pith.science/pith/N3FM7D3NI5S4XDLFOWLG3YIU43","download_json":"https://pith.science/pith/N3FM7D3NI5S4XDLFOWLG3YIU43.json","view_paper":"https://pith.science/paper/N3FM7D3N","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.02069&json=true","fetch_graph":"https://pith.science/api/pith-number/N3FM7D3NI5S4XDLFOWLG3YIU43/graph.json","fetch_events":"https://pith.science/api/pith-number/N3FM7D3NI5S4XDLFOWLG3YIU43/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/N3FM7D3NI5S4XDLFOWLG3YIU43/action/timestamp_anchor","attest_storage":"https://pith.science/pith/N3FM7D3NI5S4XDLFOWLG3YIU43/action/storage_attestation","attest_author":"https://pith.science/pith/N3FM7D3NI5S4XDLFOWLG3YIU43/action/author_attestation","sign_citation":"https://pith.science/pith/N3FM7D3NI5S4XDLFOWLG3YIU43/action/citation_signature","submit_replication":"https://pith.science/pith/N3FM7D3NI5S4XDLFOWLG3YIU43/action/replication_record"}},"created_at":"2026-07-05T10:09:23.289928+00:00","updated_at":"2026-07-05T10:09:23.289928+00:00"}