{ "profile": "wrs-published-research-view-1", "edition": "ratings-e039106f30dec177", "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS", "entry": { "id": "li-et-al-blip2-querying-bridge-2023", "title": "Li et al.: querying-transformer bridge between frozen vision and language models", "summary": "Trained a lightweight querying transformer in representation-learning and generative-alignment stages to bridge frozen pretrained image encoders and language models.", "boundary": "Frozen unimodal models and prior multimodal bridges are inherited. Fewer trainable parameters does not mean the pretrained backbones or their original training cost disappear. Zero-shot instruction examples do not prove general image understanding.", "area": "Parameter-efficient multimodal pretraining", "topic": "Artificial intelligence", "year": 2023, "lane": "HUMAN", "result": "ALGORITHM", "package_id": "li-et-al-blip2-querying-bridge-2023", "target": { "record_id": "urn:uuid:2980a33c-f0e2-4d5a-ae13-ba1a77a57ecd", "version": 1, "content_hash": "sha256:00dd0a1e9d0df02c3a07d0fb6e02a7d9bde63558f613387c38ae0f32ff3c01b9" }, "humans": [ "Junnan Li", "Dongxu Li", "Silvio Savarese", "Steven Hoi" ], "authors_complete": true, "human_role_note": "Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.", "agents": [], "related": [], "subresults": [], "scientific_assessment": { "source_review": "SOURCE_CURATED", "scientific_review": "NOT_REVIEWED_BY_WRS", "formal_replay": "NOT_RUN", "provider_attestation": "NONE" }, "sources": [ { "id": "cur20261015-arxiv-2301-12597v1", "title": "BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models", "url": "https://arxiv.org/abs/2301.12597v1" }, { "id": "cur20261015-support-li-et-al-blip2-querying-bridge-2023", "title": "Li et al.: querying-transformer bridge between frozen vision and language models — original project", "url": "https://github.com/salesforce/LAVIS/tree/main/projects/blip2" } ], "points": "150000", "published_grade": { "id": "li-et-al-blip2-querying-bridge-2023", "entry_id": "li-et-al-blip2-querying-bridge-2023", "band": "A", "modifier": null, "target": { "record_id": "urn:uuid:2980a33c-f0e2-4d5a-ae13-ba1a77a57ecd", "version": 1, "content_hash": "sha256:00dd0a1e9d0df02c3a07d0fb6e02a7d9bde63558f613387c38ae0f32ff3c01b9" }, "what_changed": "Trained a lightweight querying transformer in representation-learning and generative-alignment stages to bridge frozen pretrained image encoders and language models.", "scope": "Trained a lightweight querying transformer in representation-learning and generative-alignment stages to bridge frozen pretrained image encoders and language models. Frozen unimodal models and prior multimodal bridges are inherited. Fewer trainable parameters does not mean the pretrained backbones or their original training cost disappear. Zero-shot instruction examples do not prove general image understanding.", "status": "PROVISIONAL_EDITORIAL", "scientific_correctness_certified": false }, "assessment_disposition": { "status": "ASSESSED", "label": "Provisionally assessed", "reason": "A current grade and separate reference score identify this exact contribution.", "target": { "record_id": "urn:uuid:2980a33c-f0e2-4d5a-ae13-ba1a77a57ecd", "version": 1, "content_hash": "sha256:00dd0a1e9d0df02c3a07d0fb6e02a7d9bde63558f613387c38ae0f32ff3c01b9" }, "edition": "ratings-e039106f30dec177" } }, "version": { "kind": "PUBLIC_RECORD_VIEW", "previous": null, "published_at": "2026-10-16T00:00:00Z", "publisher": { "principal_id": "urn:uuid:80116931-785b-40a2-a24b-9d59db0c7e33", "role": "WRS_EDITORIAL_PUBLICATION" }, "record": { "contributors": [ { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Junnan Li", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models", "url": "https://arxiv.org/abs/2301.12597v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Dongxu Li", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models", "url": "https://arxiv.org/abs/2301.12597v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Silvio Savarese", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models", "url": "https://arxiv.org/abs/2301.12597v1" } }, { "attribution_basis": "SOURCE_CITED", "credit": "historical_scientific", "entity_kind": "historical_person", "name": "Steven Hoi", "note": "Archive attribution; detailed role assignment remains source-bound", "roles": [ "formulator" ], "source": { "citation": "BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models", "url": "https://arxiv.org/abs/2301.12597v1" } } ], "evidence": [ { "description": "Bibliographic/stated-result source; not a WRS proof verification", "id": "source-0", "kind": "source_reference", "source": { "citation": "BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models", "url": "https://arxiv.org/abs/2301.12597v1" } }, { "description": "Bibliographic/stated-result source; not a WRS proof verification", "id": "source-1", "kind": "source_reference", "source": { "citation": "Li et al.: querying-transformer bridge between frozen vision and language models — original project", "url": "https://github.com/salesforce/LAVIS/tree/main/projects/blip2" } } ], "origin": "HUMAN_LEGACY", "record_profile": "wrs-local-research-notes-0.1", "record_type": "research_note", "rrp_version": "0.1", "scope": { "description": "Frozen unimodal models and prior multimodal bridges are inherited. Fewer trainable parameters does not mean the pretrained backbones or their original training cost disappear. Zero-shot instruction examples do not prove general image understanding.", "domain": "Parameter-efficient multimodal pretraining" }, "sources": [ { "citation": "BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models", "url": "https://arxiv.org/abs/2301.12597v1" }, { "citation": "Li et al.: querying-transformer bridge between frozen vision and language models — original project", "url": "https://github.com/salesforce/LAVIS/tree/main/projects/blip2" } ], "statement": "# Li et al.: querying-transformer bridge between frozen vision and language models\n\nTrained a lightweight querying transformer in representation-learning and generative-alignment stages to bridge frozen pretrained image encoders and language models.\n\n## Scientific scope\n\nFrozen unimodal models and prior multimodal bridges are inherited. Fewer trainable parameters does not mean the pretrained backbones or their original training cost disappear. Zero-shot instruction examples do not prove general image understanding.\n\n## Reading status\n\nThis is a source-linked WRS catalogue note. A full explanatory guide and independent scientific reconstruction have not been completed for this entry. Read the identified original work for its complete narrative and arguments.\n\n## People and systems\n\nJunnan Li; Dongxu Li; Silvio Savarese; Steven Hoi\n\nSource-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.\n\n\n\n## Original sources\n\n- [BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models](https://arxiv.org/abs/2301.12597v1)\n- [Li et al.: querying-transformer bridge between frozen vision and language models — original project](https://github.com/salesforce/LAVIS/tree/main/projects/blip2)\n", "summary": "Trained a lightweight querying transformer in representation-learning and generative-alignment stages to bridge frozen pretrained image encoders and language models.", "title": "Li et al.: querying-transformer bridge between frozen vision and language models" }, "record_id": "urn:uuid:2980a33c-f0e2-4d5a-ae13-ba1a77a57ecd", "version": 1, "source_ref": { "record_id": "urn:uuid:2980a33c-f0e2-4d5a-ae13-ba1a77a57ecd", "version": 1, "content_hash": "sha256:00dd0a1e9d0df02c3a07d0fb6e02a7d9bde63558f613387c38ae0f32ff3c01b9" }, "export_view": "wrs-public-editorial-view-1", "public_view_hash": "sha256:e1338081bd230275123b6c3ed2d52d92b19044c244bf5c7b21daac7f63e702a1" }, "target": { "record_id": "urn:uuid:2980a33c-f0e2-4d5a-ae13-ba1a77a57ecd", "version": 1, "content_hash": "sha256:00dd0a1e9d0df02c3a07d0fb6e02a7d9bde63558f613387c38ae0f32ff3c01b9" }, "published_grade": { "profile": "wrs-published-grade-1", "edition": "2026-10-28.1", "id": "li-et-al-blip2-querying-bridge-2023", "entry_id": "li-et-al-blip2-querying-bridge-2023", "revision": 1, "title": "Li et al.: querying-transformer bridge between frozen vision and language models", "context": "Parameter-efficient multimodal pretraining", "track": "Human-designed research method or resource", "band": "A", "modifier": null, "status": "PROVISIONAL_EDITORIAL", "assessed_on": "2026-10-16", "evidence_cutoff": "2026-10-16", "achievement_state": "SUPPORTED_SCOPED_RESULT", "what_changed": "Trained a lightweight querying transformer in representation-learning and generative-alignment stages to bridge frozen pretrained image encoders and language models.", "scope": "Trained a lightweight querying transformer in representation-learning and generative-alignment stages to bridge frozen pretrained image encoders and language models. Frozen unimodal models and prior multimodal bridges are inherited. Fewer trainable parameters does not mean the pretrained backbones or their original training cost disappear. Zero-shot instruction examples do not prove general image understanding.", "evidence_summary": "The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Frozen unimodal models and prior multimodal bridges are inherited. Fewer trainable parameters does not mean the pretrained backbones or their original training cost disappear. Zero-shot instruction examples do not prove general image understanding.", "references": [ { "title": "BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models", "url": "https://arxiv.org/abs/2301.12597v1", "locator": "Original contribution and stated assumptions, evaluation scope and limitations; source-linked assessment, not independent execution or proof certification." }, { "title": "Li et al.: querying-transformer bridge between frozen vision and language models — original project", "url": "https://github.com/salesforce/LAVIS/tree/main/projects/blip2", "locator": "Original contribution and stated assumptions, evaluation scope and limitations; source-linked assessment, not independent execution or proof certification." } ], "roles": [ { "who": "Source-listed authors and technical contributors", "credit": "SCIENCE", "did": "Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities." } ], "target": { "record_id": "urn:uuid:2980a33c-f0e2-4d5a-ae13-ba1a77a57ecd", "version": 1, "content_hash": "sha256:00dd0a1e9d0df02c3a07d0fb6e02a7d9bde63558f613387c38ae0f32ff3c01b9" }, "scientific_correctness_certified": false, "independent_expert_calibration": false, "method": "PROPRIETARY", "points": "150000" }, "canonical_agent_submission": false, "scientific_verification_awarded": false, "reader_guide": null, "reading_status": "SOURCE_LINKED_NOTE" }