{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:2IWC63FCLPYRSVSEREZ2PCOAPF","short_pith_number":"pith:2IWC63FC","schema_version":"1.0","canonical_sha256":"d22c2f6ca25bf11956448933a789c0795d25f133c9345165643b58d6face2c18","source":{"kind":"arxiv","id":"2201.07520","version":1},"attestation_state":"computed","paper":{"title":"CM3: A Causal Masked Multimodal Model of the Internet","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Armen Aghajanyan, Bernie Huang, Candace Ross, Dmytro Okhonko, Gargi Ghosh, Hu Xu, Luke Zettlemoyer, Mandar Joshi, Mike Lewis, Naman Goyal, Vladimir Karpukhin","submitted_at":"2022-01-19T10:45:38Z","abstract_excerpt":"We introduce CM3, a family of causally masked generative models trained over a large corpus of structured multi-modal documents that can contain both text and image tokens. Our new causally masked approach generates tokens left to right while also masking out a small number of long token spans that are generated at the end of the string, instead of their original positions. The casual masking object provides a type of hybrid of the more common causal and masked language models, by enabling full generative modeling while also providing bidirectional context when generating the masked spans. We "},"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":"2201.07520","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-01-19T10:45:38Z","cross_cats_sorted":[],"title_canon_sha256":"f6b1054838d0a834b046022675637ef9487507ec45124ba452b35132bed56fd3","abstract_canon_sha256":"9db526d01bfafec71790c4f12124adb20dbc9a50951a2e6e76479215e24274a3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:49:47.617392Z","signature_b64":"a0FZCIREHOT76amAghnZ+EZ8zVQ2Sbm/XpvcjO376J9SEn1ngAUxzj43HjvraJkHQtQxGqoWDXT8NiPIM88KCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d22c2f6ca25bf11956448933a789c0795d25f133c9345165643b58d6face2c18","last_reissued_at":"2026-07-05T03:49:47.616983Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:49:47.616983Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CM3: A Causal Masked Multimodal Model of the Internet","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Armen Aghajanyan, Bernie Huang, Candace Ross, Dmytro Okhonko, Gargi Ghosh, Hu Xu, Luke Zettlemoyer, Mandar Joshi, Mike Lewis, Naman Goyal, Vladimir Karpukhin","submitted_at":"2022-01-19T10:45:38Z","abstract_excerpt":"We introduce CM3, a family of causally masked generative models trained over a large corpus of structured multi-modal documents that can contain both text and image tokens. Our new causally masked approach generates tokens left to right while also masking out a small number of long token spans that are generated at the end of the string, instead of their original positions. The casual masking object provides a type of hybrid of the more common causal and masked language models, by enabling full generative modeling while also providing bidirectional context when generating the masked spans. We "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2201.07520","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/2201.07520/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":"2201.07520","created_at":"2026-07-05T03:49:47.617041+00:00"},{"alias_kind":"arxiv_version","alias_value":"2201.07520v1","created_at":"2026-07-05T03:49:47.617041+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2201.07520","created_at":"2026-07-05T03:49:47.617041+00:00"},{"alias_kind":"pith_short_12","alias_value":"2IWC63FCLPYR","created_at":"2026-07-05T03:49:47.617041+00:00"},{"alias_kind":"pith_short_16","alias_value":"2IWC63FCLPYRSVSE","created_at":"2026-07-05T03:49:47.617041+00:00"},{"alias_kind":"pith_short_8","alias_value":"2IWC63FC","created_at":"2026-07-05T03:49:47.617041+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":14,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.17118","citing_title":"Differentiable Optimization Layers for Guaranteed Fairness in Deep Learning","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2411.04996","citing_title":"Mixture-of-Transformers: A Sparse and Scalable Architecture for Multi-Modal Foundation Models","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2207.14255","citing_title":"Efficient Training of Language Models to Fill in the Middle","ref_index":95,"is_internal_anchor":false},{"citing_arxiv_id":"2412.14164","citing_title":"MetaMorph: Multimodal Understanding and Generation via Instruction Tuning","ref_index":276,"is_internal_anchor":false},{"citing_arxiv_id":"2309.15505","citing_title":"Finite Scalar Quantization: VQ-VAE Made Simple","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2204.05999","citing_title":"InCoder: A Generative Model for Code Infilling and Synthesis","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2302.14045","citing_title":"Language Is Not All You Need: Aligning Perception with Language Models","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2401.01335","citing_title":"Self-Play Fine-Tuning Converts Weak Language Models to Strong Language Models","ref_index":290,"is_internal_anchor":false},{"citing_arxiv_id":"2308.01390","citing_title":"OpenFlamingo: An Open-Source Framework for Training Large Autoregressive Vision-Language Models","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2306.14824","citing_title":"Kosmos-2: Grounding Multimodal Large Language Models to the World","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2204.14198","citing_title":"Flamingo: a Visual Language Model for Few-Shot Learning","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2407.07726","citing_title":"PaliGemma: A versatile 3B VLM for transfer","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2405.09818","citing_title":"Chameleon: Mixed-Modal Early-Fusion Foundation Models","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2204.06125","citing_title":"Hierarchical Text-Conditional Image Generation with CLIP Latents","ref_index":1,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2IWC63FCLPYRSVSEREZ2PCOAPF","json":"https://pith.science/pith/2IWC63FCLPYRSVSEREZ2PCOAPF.json","graph_json":"https://pith.science/api/pith-number/2IWC63FCLPYRSVSEREZ2PCOAPF/graph.json","events_json":"https://pith.science/api/pith-number/2IWC63FCLPYRSVSEREZ2PCOAPF/events.json","paper":"https://pith.science/paper/2IWC63FC"},"agent_actions":{"view_html":"https://pith.science/pith/2IWC63FCLPYRSVSEREZ2PCOAPF","download_json":"https://pith.science/pith/2IWC63FCLPYRSVSEREZ2PCOAPF.json","view_paper":"https://pith.science/paper/2IWC63FC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2201.07520&json=true","fetch_graph":"https://pith.science/api/pith-number/2IWC63FCLPYRSVSEREZ2PCOAPF/graph.json","fetch_events":"https://pith.science/api/pith-number/2IWC63FCLPYRSVSEREZ2PCOAPF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2IWC63FCLPYRSVSEREZ2PCOAPF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2IWC63FCLPYRSVSEREZ2PCOAPF/action/storage_attestation","attest_author":"https://pith.science/pith/2IWC63FCLPYRSVSEREZ2PCOAPF/action/author_attestation","sign_citation":"https://pith.science/pith/2IWC63FCLPYRSVSEREZ2PCOAPF/action/citation_signature","submit_replication":"https://pith.science/pith/2IWC63FCLPYRSVSEREZ2PCOAPF/action/replication_record"}},"created_at":"2026-07-05T03:49:47.617041+00:00","updated_at":"2026-07-05T03:49:47.617041+00:00"}