{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:6PNAOWC7OR6QWOU2DFUI6M5K3W","short_pith_number":"pith:6PNAOWC7","schema_version":"1.0","canonical_sha256":"f3da07585f747d0b3a9a19688f33aaddb0cf0e70888df4d6bc53d87f74ad8283","source":{"kind":"arxiv","id":"2312.12661","version":1},"attestation_state":"computed","paper":{"title":"Misalign, Contrast then Distill: Rethinking Misalignments in Language-Image Pretraining","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bumsoo Kim, Jinhyung Kim, Seung Hwan Kim, Yeonsik Jo","submitted_at":"2023-12-19T23:22:47Z","abstract_excerpt":"Contrastive Language-Image Pretraining has emerged as a prominent approach for training vision and text encoders with uncurated image-text pairs from the web. To enhance data-efficiency, recent efforts have introduced additional supervision terms that involve random-augmented views of the image. However, since the image augmentation process is unaware of its text counterpart, this procedure could cause various degrees of image-text misalignments during training. Prior methods either disregarded this discrepancy or introduced external models to mitigate the impact of misalignments during traini"},"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":"2312.12661","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2023-12-19T23:22:47Z","cross_cats_sorted":[],"title_canon_sha256":"6c2d173eabc2045d402c21c056fac02edc5126f9b4a8fe1e11e0bee325c6c534","abstract_canon_sha256":"0dda763031da337cd87e9e6c77ee05dbaa74e5b336385327843fe65de533ee35"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:26:24.670219Z","signature_b64":"74MXC7j3M1ugklnFWn6+mspTcReGcuMb0fK9uW8Rgf/BOo+ldzzwPpvPqx3LljfmhVDY8tn7d5xLoeDH+BsBCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f3da07585f747d0b3a9a19688f33aaddb0cf0e70888df4d6bc53d87f74ad8283","last_reissued_at":"2026-07-05T07:26:24.669738Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:26:24.669738Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Misalign, Contrast then Distill: Rethinking Misalignments in Language-Image Pretraining","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bumsoo Kim, Jinhyung Kim, Seung Hwan Kim, Yeonsik Jo","submitted_at":"2023-12-19T23:22:47Z","abstract_excerpt":"Contrastive Language-Image Pretraining has emerged as a prominent approach for training vision and text encoders with uncurated image-text pairs from the web. To enhance data-efficiency, recent efforts have introduced additional supervision terms that involve random-augmented views of the image. However, since the image augmentation process is unaware of its text counterpart, this procedure could cause various degrees of image-text misalignments during training. Prior methods either disregarded this discrepancy or introduced external models to mitigate the impact of misalignments during traini"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2312.12661","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/2312.12661/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":"2312.12661","created_at":"2026-07-05T07:26:24.669799+00:00"},{"alias_kind":"arxiv_version","alias_value":"2312.12661v1","created_at":"2026-07-05T07:26:24.669799+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2312.12661","created_at":"2026-07-05T07:26:24.669799+00:00"},{"alias_kind":"pith_short_12","alias_value":"6PNAOWC7OR6Q","created_at":"2026-07-05T07:26:24.669799+00:00"},{"alias_kind":"pith_short_16","alias_value":"6PNAOWC7OR6QWOU2","created_at":"2026-07-05T07:26:24.669799+00:00"},{"alias_kind":"pith_short_8","alias_value":"6PNAOWC7","created_at":"2026-07-05T07:26:24.669799+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2502.07466","citing_title":"Less is More: Masking Elements in Image Condition Features Avoids Content Leakages in Style Transfer Diffusion Models","ref_index":17,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6PNAOWC7OR6QWOU2DFUI6M5K3W","json":"https://pith.science/pith/6PNAOWC7OR6QWOU2DFUI6M5K3W.json","graph_json":"https://pith.science/api/pith-number/6PNAOWC7OR6QWOU2DFUI6M5K3W/graph.json","events_json":"https://pith.science/api/pith-number/6PNAOWC7OR6QWOU2DFUI6M5K3W/events.json","paper":"https://pith.science/paper/6PNAOWC7"},"agent_actions":{"view_html":"https://pith.science/pith/6PNAOWC7OR6QWOU2DFUI6M5K3W","download_json":"https://pith.science/pith/6PNAOWC7OR6QWOU2DFUI6M5K3W.json","view_paper":"https://pith.science/paper/6PNAOWC7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2312.12661&json=true","fetch_graph":"https://pith.science/api/pith-number/6PNAOWC7OR6QWOU2DFUI6M5K3W/graph.json","fetch_events":"https://pith.science/api/pith-number/6PNAOWC7OR6QWOU2DFUI6M5K3W/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6PNAOWC7OR6QWOU2DFUI6M5K3W/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6PNAOWC7OR6QWOU2DFUI6M5K3W/action/storage_attestation","attest_author":"https://pith.science/pith/6PNAOWC7OR6QWOU2DFUI6M5K3W/action/author_attestation","sign_citation":"https://pith.science/pith/6PNAOWC7OR6QWOU2DFUI6M5K3W/action/citation_signature","submit_replication":"https://pith.science/pith/6PNAOWC7OR6QWOU2DFUI6M5K3W/action/replication_record"}},"created_at":"2026-07-05T07:26:24.669799+00:00","updated_at":"2026-07-05T07:26:24.669799+00:00"}