{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:7P3CY75E6LO4QPGV43AYZADK6W","short_pith_number":"pith:7P3CY75E","schema_version":"1.0","canonical_sha256":"fbf62c7fa4f2ddc83cd5e6c18c806af59e29681003455b8470a93444e105cf6a","source":{"kind":"arxiv","id":"2506.02020","version":1},"attestation_state":"computed","paper":{"title":"Improve Multi-Modal Embedding Learning via Explicit Hard Negative Gradient Amplifying","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Dian Li, Gang Liu, Youze Xue","submitted_at":"2025-05-28T11:18:19Z","abstract_excerpt":"With the rapid advancement of multi-modal large language models (MLLMs) in recent years, the foundational Contrastive Language-Image Pretraining (CLIP) framework has been successfully extended to MLLMs, enabling more powerful and universal multi-modal embeddings for a wide range of retrieval tasks. Despite these developments, the core contrastive learning paradigm remains largely unchanged from CLIP-style models to MLLMs. Within this framework, the effective mining of hard negative samples continues to be a critical factor for enhancing performance. Prior works have introduced both offline and"},"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":"2506.02020","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2025-05-28T11:18:19Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"84f31a533f7e6eee7cb037f0a629bc7acd2da8d366bb1a74f3f82fc689a7e072","abstract_canon_sha256":"0167bf45c02b3ed864dd1f39d7eba7907cfb0d7b2c5799f1ed4085acfa365baf"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:14:31.038685Z","signature_b64":"URoAFbaOGith7RE3jIr9eIivX47hJ6m91N+ufmX9X3zpMbodCrVHICPAMh9yioRy5h3J1OwY7Eg3O2UgKI4xBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fbf62c7fa4f2ddc83cd5e6c18c806af59e29681003455b8470a93444e105cf6a","last_reissued_at":"2026-07-05T11:14:31.038237Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:14:31.038237Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Improve Multi-Modal Embedding Learning via Explicit Hard Negative Gradient Amplifying","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Dian Li, Gang Liu, Youze Xue","submitted_at":"2025-05-28T11:18:19Z","abstract_excerpt":"With the rapid advancement of multi-modal large language models (MLLMs) in recent years, the foundational Contrastive Language-Image Pretraining (CLIP) framework has been successfully extended to MLLMs, enabling more powerful and universal multi-modal embeddings for a wide range of retrieval tasks. Despite these developments, the core contrastive learning paradigm remains largely unchanged from CLIP-style models to MLLMs. Within this framework, the effective mining of hard negative samples continues to be a critical factor for enhancing performance. Prior works have introduced both offline and"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.02020","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/2506.02020/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":"2506.02020","created_at":"2026-07-05T11:14:31.038293+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.02020v1","created_at":"2026-07-05T11:14:31.038293+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.02020","created_at":"2026-07-05T11:14:31.038293+00:00"},{"alias_kind":"pith_short_12","alias_value":"7P3CY75E6LO4","created_at":"2026-07-05T11:14:31.038293+00:00"},{"alias_kind":"pith_short_16","alias_value":"7P3CY75E6LO4QPGV","created_at":"2026-07-05T11:14:31.038293+00:00"},{"alias_kind":"pith_short_8","alias_value":"7P3CY75E","created_at":"2026-07-05T11:14:31.038293+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.04976","citing_title":"Tencent Advertising Algorithm Challenge 2025: All-Modality Generative Recommendation","ref_index":61,"is_internal_anchor":false},{"citing_arxiv_id":"2605.02378","citing_title":"Enhancing Multimodal In-Context Learning via Inductive-Deductive Reasoning","ref_index":46,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7P3CY75E6LO4QPGV43AYZADK6W","json":"https://pith.science/pith/7P3CY75E6LO4QPGV43AYZADK6W.json","graph_json":"https://pith.science/api/pith-number/7P3CY75E6LO4QPGV43AYZADK6W/graph.json","events_json":"https://pith.science/api/pith-number/7P3CY75E6LO4QPGV43AYZADK6W/events.json","paper":"https://pith.science/paper/7P3CY75E"},"agent_actions":{"view_html":"https://pith.science/pith/7P3CY75E6LO4QPGV43AYZADK6W","download_json":"https://pith.science/pith/7P3CY75E6LO4QPGV43AYZADK6W.json","view_paper":"https://pith.science/paper/7P3CY75E","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.02020&json=true","fetch_graph":"https://pith.science/api/pith-number/7P3CY75E6LO4QPGV43AYZADK6W/graph.json","fetch_events":"https://pith.science/api/pith-number/7P3CY75E6LO4QPGV43AYZADK6W/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7P3CY75E6LO4QPGV43AYZADK6W/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7P3CY75E6LO4QPGV43AYZADK6W/action/storage_attestation","attest_author":"https://pith.science/pith/7P3CY75E6LO4QPGV43AYZADK6W/action/author_attestation","sign_citation":"https://pith.science/pith/7P3CY75E6LO4QPGV43AYZADK6W/action/citation_signature","submit_replication":"https://pith.science/pith/7P3CY75E6LO4QPGV43AYZADK6W/action/replication_record"}},"created_at":"2026-07-05T11:14:31.038293+00:00","updated_at":"2026-07-05T11:14:31.038293+00:00"}