{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:3HK3ZCIWWWFKPBCWU3CSSYAE2F","short_pith_number":"pith:3HK3ZCIW","schema_version":"1.0","canonical_sha256":"d9d5bc8916b58aa78456a6c5296004d174ae8cff09bb98fd550eb2212409950a","source":{"kind":"arxiv","id":"2310.09930","version":1},"attestation_state":"computed","paper":{"title":"FiLM: Fill-in Language Models for Any-Order Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Hao Peng, Ruoqi Shen, Tianxiao Shen, Yao Fu, Yejin Choi, Zaid Harchaoui","submitted_at":"2023-10-15T19:37:39Z","abstract_excerpt":"Language models have become the backbone of today's AI systems. However, their predominant left-to-right generation limits the use of bidirectional context, which is essential for tasks that involve filling text in the middle. We propose the Fill-in Language Model (FiLM), a new language modeling approach that allows for flexible generation at any position without adhering to a specific generation order. Its training extends the masked language modeling objective by adopting varying mask probabilities sampled from the Beta distribution to enhance the generative capabilities of FiLM. During infe"},"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":"2310.09930","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-10-15T19:37:39Z","cross_cats_sorted":[],"title_canon_sha256":"b8524e9704065b8d4d98cee6d05c46941612d5add5d66342e073f2b29f2957df","abstract_canon_sha256":"03709aebc5bacb3f425061a8697644e6615de3fa98c85c6f57e79ec6c4629c74"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:01:12.760056Z","signature_b64":"MW9csofJc4c1Ih4KRQuhKba8g6mwfvBA+56Tt8Mdb4mUUzuY5/JpjyFzfXyYwTmxajZca6clUZZcQ1fuXm6GAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d9d5bc8916b58aa78456a6c5296004d174ae8cff09bb98fd550eb2212409950a","last_reissued_at":"2026-07-05T07:01:12.759571Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:01:12.759571Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"FiLM: Fill-in Language Models for Any-Order Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Hao Peng, Ruoqi Shen, Tianxiao Shen, Yao Fu, Yejin Choi, Zaid Harchaoui","submitted_at":"2023-10-15T19:37:39Z","abstract_excerpt":"Language models have become the backbone of today's AI systems. However, their predominant left-to-right generation limits the use of bidirectional context, which is essential for tasks that involve filling text in the middle. We propose the Fill-in Language Model (FiLM), a new language modeling approach that allows for flexible generation at any position without adhering to a specific generation order. Its training extends the masked language modeling objective by adopting varying mask probabilities sampled from the Beta distribution to enhance the generative capabilities of FiLM. During infe"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.09930","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/2310.09930/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":"2310.09930","created_at":"2026-07-05T07:01:12.759634+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.09930v1","created_at":"2026-07-05T07:01:12.759634+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.09930","created_at":"2026-07-05T07:01:12.759634+00:00"},{"alias_kind":"pith_short_12","alias_value":"3HK3ZCIWWWFK","created_at":"2026-07-05T07:01:12.759634+00:00"},{"alias_kind":"pith_short_16","alias_value":"3HK3ZCIWWWFKPBCW","created_at":"2026-07-05T07:01:12.759634+00:00"},{"alias_kind":"pith_short_8","alias_value":"3HK3ZCIW","created_at":"2026-07-05T07:01:12.759634+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.01775","citing_title":"Set Diffusion: Interpolating Token Orderings Between Autoregression and Diffusion for Fast and Flexible Decoding","ref_index":141,"is_internal_anchor":false},{"citing_arxiv_id":"2410.17891","citing_title":"Scaling Diffusion Language Models via Adaptation from Autoregressive Models","ref_index":176,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3HK3ZCIWWWFKPBCWU3CSSYAE2F","json":"https://pith.science/pith/3HK3ZCIWWWFKPBCWU3CSSYAE2F.json","graph_json":"https://pith.science/api/pith-number/3HK3ZCIWWWFKPBCWU3CSSYAE2F/graph.json","events_json":"https://pith.science/api/pith-number/3HK3ZCIWWWFKPBCWU3CSSYAE2F/events.json","paper":"https://pith.science/paper/3HK3ZCIW"},"agent_actions":{"view_html":"https://pith.science/pith/3HK3ZCIWWWFKPBCWU3CSSYAE2F","download_json":"https://pith.science/pith/3HK3ZCIWWWFKPBCWU3CSSYAE2F.json","view_paper":"https://pith.science/paper/3HK3ZCIW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.09930&json=true","fetch_graph":"https://pith.science/api/pith-number/3HK3ZCIWWWFKPBCWU3CSSYAE2F/graph.json","fetch_events":"https://pith.science/api/pith-number/3HK3ZCIWWWFKPBCWU3CSSYAE2F/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3HK3ZCIWWWFKPBCWU3CSSYAE2F/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3HK3ZCIWWWFKPBCWU3CSSYAE2F/action/storage_attestation","attest_author":"https://pith.science/pith/3HK3ZCIWWWFKPBCWU3CSSYAE2F/action/author_attestation","sign_citation":"https://pith.science/pith/3HK3ZCIWWWFKPBCWU3CSSYAE2F/action/citation_signature","submit_replication":"https://pith.science/pith/3HK3ZCIWWWFKPBCWU3CSSYAE2F/action/replication_record"}},"created_at":"2026-07-05T07:01:12.759634+00:00","updated_at":"2026-07-05T07:01:12.759634+00:00"}