{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:IAZSHC2CEOVJS5LFPG2ACLALBB","short_pith_number":"pith:IAZSHC2C","schema_version":"1.0","canonical_sha256":"4033238b4223aa99756579b4012c0b087ef2121d56f6c9c345a18709683a0609","source":{"kind":"arxiv","id":"2401.03321","version":2},"attestation_state":"computed","paper":{"title":"PIXAR: Auto-Regressive Language Modeling in Pixel Space","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Alessandro Suglia, Antonio Vergari, Xiyang Liao, Yintao Tai","submitted_at":"2024-01-06T22:49:38Z","abstract_excerpt":"Recent work showed the possibility of building open-vocabulary large language models (LLMs) that directly operate on pixel representations. These models are implemented as autoencoders that reconstruct masked patches of rendered text. However, these pixel-based LLMs are limited to discriminative tasks (e.g., classification) and, similar to BERT, cannot be used to generate text. Therefore, they cannot be used for generative tasks such as free-form question answering. In this work, we introduce PIXAR, the first pixel-based autoregressive LLM that performs text generation. Consisting of only a de"},"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":"2401.03321","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2024-01-06T22:49:38Z","cross_cats_sorted":[],"title_canon_sha256":"116acf5c8e86ca342bfc6e9b68d7e0a213ef1043dd459235cf3e67ded5134f23","abstract_canon_sha256":"15fed9cdec205952b193885672ba07c81d187e36d77f407888593a2b86873241"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:48:51.619127Z","signature_b64":"5Uv20MPVeAG7ULAvdzHyIkpom/JLUrkVqkZbulGEx7f3AgtIXk7VVEnvITcU/JWmh+aZ8YbtzhjSA/dGuJ15Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4033238b4223aa99756579b4012c0b087ef2121d56f6c9c345a18709683a0609","last_reissued_at":"2026-07-05T07:48:51.618690Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:48:51.618690Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"PIXAR: Auto-Regressive Language Modeling in Pixel Space","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Alessandro Suglia, Antonio Vergari, Xiyang Liao, Yintao Tai","submitted_at":"2024-01-06T22:49:38Z","abstract_excerpt":"Recent work showed the possibility of building open-vocabulary large language models (LLMs) that directly operate on pixel representations. These models are implemented as autoencoders that reconstruct masked patches of rendered text. However, these pixel-based LLMs are limited to discriminative tasks (e.g., classification) and, similar to BERT, cannot be used to generate text. Therefore, they cannot be used for generative tasks such as free-form question answering. In this work, we introduce PIXAR, the first pixel-based autoregressive LLM that performs text generation. Consisting of only a de"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.03321","kind":"arxiv","version":2},"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/2401.03321/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":"2401.03321","created_at":"2026-07-05T07:48:51.618749+00:00"},{"alias_kind":"arxiv_version","alias_value":"2401.03321v2","created_at":"2026-07-05T07:48:51.618749+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.03321","created_at":"2026-07-05T07:48:51.618749+00:00"},{"alias_kind":"pith_short_12","alias_value":"IAZSHC2CEOVJ","created_at":"2026-07-05T07:48:51.618749+00:00"},{"alias_kind":"pith_short_16","alias_value":"IAZSHC2CEOVJS5LF","created_at":"2026-07-05T07:48:51.618749+00:00"},{"alias_kind":"pith_short_8","alias_value":"IAZSHC2C","created_at":"2026-07-05T07:48:51.618749+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.11575","citing_title":"MIXAR: Scaling Autoregressive Pixel-based Language Models to Multiple Languages and Scripts","ref_index":27,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IAZSHC2CEOVJS5LFPG2ACLALBB","json":"https://pith.science/pith/IAZSHC2CEOVJS5LFPG2ACLALBB.json","graph_json":"https://pith.science/api/pith-number/IAZSHC2CEOVJS5LFPG2ACLALBB/graph.json","events_json":"https://pith.science/api/pith-number/IAZSHC2CEOVJS5LFPG2ACLALBB/events.json","paper":"https://pith.science/paper/IAZSHC2C"},"agent_actions":{"view_html":"https://pith.science/pith/IAZSHC2CEOVJS5LFPG2ACLALBB","download_json":"https://pith.science/pith/IAZSHC2CEOVJS5LFPG2ACLALBB.json","view_paper":"https://pith.science/paper/IAZSHC2C","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2401.03321&json=true","fetch_graph":"https://pith.science/api/pith-number/IAZSHC2CEOVJS5LFPG2ACLALBB/graph.json","fetch_events":"https://pith.science/api/pith-number/IAZSHC2CEOVJS5LFPG2ACLALBB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IAZSHC2CEOVJS5LFPG2ACLALBB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IAZSHC2CEOVJS5LFPG2ACLALBB/action/storage_attestation","attest_author":"https://pith.science/pith/IAZSHC2CEOVJS5LFPG2ACLALBB/action/author_attestation","sign_citation":"https://pith.science/pith/IAZSHC2CEOVJS5LFPG2ACLALBB/action/citation_signature","submit_replication":"https://pith.science/pith/IAZSHC2CEOVJS5LFPG2ACLALBB/action/replication_record"}},"created_at":"2026-07-05T07:48:51.618749+00:00","updated_at":"2026-07-05T07:48:51.618749+00:00"}