{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:CU7RBPWBK4CQEILSCW7XVACKSY","short_pith_number":"pith:CU7RBPWB","schema_version":"1.0","canonical_sha256":"153f10bec1570502217215bf7a804a96312678628318a8b2039f53ccd4e44a7d","source":{"kind":"arxiv","id":"2412.15188","version":4},"attestation_state":"computed","paper":{"title":"LMFusion: Adapting Pretrained Language Models for Multimodal Generation","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CV","cs.LG"],"primary_cat":"cs.CL","authors_text":"Chunting Zhou, Lili Yu, Luke Zettlemoyer, Weijia Shi, Weixin Liang, Xiaochuang Han, Xi Victoria Lin","submitted_at":"2024-12-19T18:56:24Z","abstract_excerpt":"We present LMFusion, a framework for empowering pretrained text-only large language models (LLMs) with multimodal generative capabilities, enabling them to understand and generate both text and images in arbitrary sequences. LMFusion leverages existing Llama-3's weights for processing texts autoregressively while introducing additional and parallel transformer modules for processing images with diffusion. During training, the data from each modality is routed to its dedicated modules: modality-specific feedforward layers, query-key-value projections, and normalization layers process each modal"},"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":"2412.15188","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2024-12-19T18:56:24Z","cross_cats_sorted":["cs.AI","cs.CV","cs.LG"],"title_canon_sha256":"4fcf04f9d183675524ebe4f8195841490ab382ee866e9184c31bf81e91a00162","abstract_canon_sha256":"dfa51f464f94b4f93b696e3a11dc4b58fff6230661dd8bbbcf1ff599e16cbd66"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:09:50.792598Z","signature_b64":"nHSFS/ohrMEHZG6//CtCcRe1WtqvyEacY55FMth36qb1evLT2kQEYygwWaXnnZQ1e/dqLeLNfBv5+eh++GqyDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"153f10bec1570502217215bf7a804a96312678628318a8b2039f53ccd4e44a7d","last_reissued_at":"2026-07-05T10:09:50.792050Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:09:50.792050Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LMFusion: Adapting Pretrained Language Models for Multimodal Generation","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CV","cs.LG"],"primary_cat":"cs.CL","authors_text":"Chunting Zhou, Lili Yu, Luke Zettlemoyer, Weijia Shi, Weixin Liang, Xiaochuang Han, Xi Victoria Lin","submitted_at":"2024-12-19T18:56:24Z","abstract_excerpt":"We present LMFusion, a framework for empowering pretrained text-only large language models (LLMs) with multimodal generative capabilities, enabling them to understand and generate both text and images in arbitrary sequences. LMFusion leverages existing Llama-3's weights for processing texts autoregressively while introducing additional and parallel transformer modules for processing images with diffusion. During training, the data from each modality is routed to its dedicated modules: modality-specific feedforward layers, query-key-value projections, and normalization layers process each modal"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.15188","kind":"arxiv","version":4},"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/2412.15188/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":"2412.15188","created_at":"2026-07-05T10:09:50.792107+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.15188v4","created_at":"2026-07-05T10:09:50.792107+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.15188","created_at":"2026-07-05T10:09:50.792107+00:00"},{"alias_kind":"pith_short_12","alias_value":"CU7RBPWBK4CQ","created_at":"2026-07-05T10:09:50.792107+00:00"},{"alias_kind":"pith_short_16","alias_value":"CU7RBPWBK4CQEILS","created_at":"2026-07-05T10:09:50.792107+00:00"},{"alias_kind":"pith_short_8","alias_value":"CU7RBPWB","created_at":"2026-07-05T10:09:50.792107+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":21,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.01858","citing_title":"Polaris: Scaling Up Instruction-Guided Image Generation Towards Millions of Personalized Style Needs","ref_index":97,"is_internal_anchor":false},{"citing_arxiv_id":"2606.31451","citing_title":"UniTac: A Unified Multimodal Model for Cross-Sensor Tactile Understanding and Generation","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18714","citing_title":"Semantic Generative Tuning for Unified Multimodal Models","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00310","citing_title":"Where to Refine, When to Stop: Rethinking Redundancy via Latent Discrepancy for Efficient Visual Autoregressive Generation","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2601.21798","citing_title":"CG-MLLM: Captioning and Generating 3D content via Multi-modal Large Language Models","ref_index":59,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15792","citing_title":"Reversing the Flow: Generation-to-Understanding Synergy in Large Multimodal Models","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18714","citing_title":"Semantic Generative Tuning for Unified Multimodal Models","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2506.18871","citing_title":"OmniGen2: Towards Instruction-Aligned Multimodal Generation","ref_index":68,"is_internal_anchor":false},{"citing_arxiv_id":"2602.12370","citing_title":"LLaMo: Scaling Pretrained Language Models for Unified Motion Understanding and Generation with Continuous Autoregressive Tokens","ref_index":58,"is_internal_anchor":false},{"citing_arxiv_id":"2503.07265","citing_title":"WISE: A World Knowledge-Informed Semantic Evaluation for Text-to-Image Generation","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2504.06256","citing_title":"Transfer between Modalities with MetaQueries","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2506.15564","citing_title":"Show-o2: Improved Native Unified Multimodal Models","ref_index":95,"is_internal_anchor":false},{"citing_arxiv_id":"2604.26341","citing_title":"SpatialFusion: Endowing Unified Image Generation with Intrinsic 3D Geometric Awareness","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2505.09568","citing_title":"BLIP3-o: A Family of Fully Open Unified Multimodal Models-Architecture, Training and Dataset","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2604.24625","citing_title":"Meta-CoT: Enhancing Granularity and Generalization in Image Editing","ref_index":52,"is_internal_anchor":false},{"citing_arxiv_id":"2604.08719","citing_title":"LMGenDrive: Bridging Multimodal Understanding and Generative World Modeling for End-to-End Driving","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08029","citing_title":"STARFlow2: Bridging Language Models and Normalizing Flows for Unified Multimodal Generation","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2505.14683","citing_title":"Emerging Properties in Unified Multimodal Pretraining","ref_index":66,"is_internal_anchor":false},{"citing_arxiv_id":"2604.13540","citing_title":"Free Lunch for Unified Multimodal Models: Enhancing Generation via Reflective Rectification with Inherent Understanding","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16552","citing_title":"Co-generation of Layout and Shape from Text via Autoregressive 3D Diffusion","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19902","citing_title":"MMCORE: MultiModal COnnection with Representation Aligned Latent Embeddings","ref_index":31,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CU7RBPWBK4CQEILSCW7XVACKSY","json":"https://pith.science/pith/CU7RBPWBK4CQEILSCW7XVACKSY.json","graph_json":"https://pith.science/api/pith-number/CU7RBPWBK4CQEILSCW7XVACKSY/graph.json","events_json":"https://pith.science/api/pith-number/CU7RBPWBK4CQEILSCW7XVACKSY/events.json","paper":"https://pith.science/paper/CU7RBPWB"},"agent_actions":{"view_html":"https://pith.science/pith/CU7RBPWBK4CQEILSCW7XVACKSY","download_json":"https://pith.science/pith/CU7RBPWBK4CQEILSCW7XVACKSY.json","view_paper":"https://pith.science/paper/CU7RBPWB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.15188&json=true","fetch_graph":"https://pith.science/api/pith-number/CU7RBPWBK4CQEILSCW7XVACKSY/graph.json","fetch_events":"https://pith.science/api/pith-number/CU7RBPWBK4CQEILSCW7XVACKSY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CU7RBPWBK4CQEILSCW7XVACKSY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CU7RBPWBK4CQEILSCW7XVACKSY/action/storage_attestation","attest_author":"https://pith.science/pith/CU7RBPWBK4CQEILSCW7XVACKSY/action/author_attestation","sign_citation":"https://pith.science/pith/CU7RBPWBK4CQEILSCW7XVACKSY/action/citation_signature","submit_replication":"https://pith.science/pith/CU7RBPWBK4CQEILSCW7XVACKSY/action/replication_record"}},"created_at":"2026-07-05T10:09:50.792107+00:00","updated_at":"2026-07-05T10:09:50.792107+00:00"}