{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:RADVO4ZGCQO7VFKWNOFW3QRZEV","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"b54915ca9ea819f2c347239d64dc926c9b8813cb1d7b177bb46ddbe7083d99a2","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2025-10-10T21:43:50Z","title_canon_sha256":"ac8e2cde2e0a5981f7ccbd503bdf9f1b6993a0e2a55da1770bad22efc1ee93e2"},"schema_version":"1.0","source":{"id":"2510.09885","kind":"arxiv","version":6}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2510.09885","created_at":"2026-06-11T01:09:19Z"},{"alias_kind":"arxiv_version","alias_value":"2510.09885v6","created_at":"2026-06-11T01:09:19Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2510.09885","created_at":"2026-06-11T01:09:19Z"},{"alias_kind":"pith_short_12","alias_value":"RADVO4ZGCQO7","created_at":"2026-06-11T01:09:19Z"},{"alias_kind":"pith_short_16","alias_value":"RADVO4ZGCQO7VFKW","created_at":"2026-06-11T01:09:19Z"},{"alias_kind":"pith_short_8","alias_value":"RADVO4ZG","created_at":"2026-06-11T01:09:19Z"}],"graph_snapshots":[{"event_id":"sha256:07d2c8b4f8cbdf6e644887ff8fccb4c5765aec1ee7176acca6c6e5457ddc5dac","target":"graph","created_at":"2026-06-11T01:09:19Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"The masked fine-tuning for arLLMs substantially improves the efficacy of knowledge injection, i.e. no paraphrase needed and resistant to the reversal curse, closing the gap between arLLMs and dLLMs."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the demasking objective alone induces the observed knowledge-injection advantage in arLLMs independent of diffusion-specific architecture details, and that the controlled experiments isolate this effect without confounding differences in model scale, data distribution, or masking implementation."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Masked fine-tuning enables autoregressive LLMs to inject new factual knowledge without paraphrases and with reversal-curse resistance, matching diffusion LLM advantages on QA tasks."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Masked fine-tuning lets autoregressive LLMs absorb new facts without paraphrases and without reversal-curse failures."}],"snapshot_sha256":"d9d6daa9402ffd87a5266e86cad8b4d6e476d3026019d589a317f4fe0cdb35bf"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2510.09885/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large language models (LLMs) are often used in environments where facts evolve, yet factual knowledge updates via fine-tuning on unstructured text often suffer from 1) reliance on compute-heavy paraphrasing augmentation and 2) the reversal curse. Recent studies show diffusion large language models (dLLMs) require fewer training samples to achieve lower loss in pre-training and are more resistant to the reversal curse, suggesting dLLMs may learn new knowledge more easily than autoregressive LLMs (arLLMs). We test this hypothesis in controlled knowledge fine-tuning experiments and find that whil","authors_text":"Ely Hahami, Haim Sompolinsky, Jingxuan Fan, Xu Pan, Ziqian Xie","cross_cats":["cs.AI"],"headline":"Masked fine-tuning lets autoregressive LLMs absorb new facts without paraphrases and without reversal-curse failures.","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2025-10-10T21:43:50Z","title":"Diffusion-Inspired Masked Fine-Tuning for Knowledge Injection in Autoregressive LLMs"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2510.09885","kind":"arxiv","version":6},"verdict":{"created_at":"2026-05-18T07:18:49.407331Z","id":"4305538a-0d9a-4d9c-855d-0ec86d924738","model_set":{"reader":"grok-4.3"},"one_line_summary":"Masked fine-tuning enables autoregressive LLMs to inject new factual knowledge without paraphrases and with reversal-curse resistance, matching diffusion LLM advantages on QA tasks.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Masked fine-tuning lets autoregressive LLMs absorb new facts without paraphrases and without reversal-curse failures.","strongest_claim":"The masked fine-tuning for arLLMs substantially improves the efficacy of knowledge injection, i.e. no paraphrase needed and resistant to the reversal curse, closing the gap between arLLMs and dLLMs.","weakest_assumption":"That the demasking objective alone induces the observed knowledge-injection advantage in arLLMs independent of diffusion-specific architecture details, and that the controlled experiments isolate this effect without confounding differences in model scale, data distribution, or masking implementation."}},"verdict_id":"4305538a-0d9a-4d9c-855d-0ec86d924738"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:6a33c8a9f42032c6e46117dfd0781a81a60e06935d9ecad48100cfc4a327efdc","target":"record","created_at":"2026-06-11T01:09:19Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"b54915ca9ea819f2c347239d64dc926c9b8813cb1d7b177bb46ddbe7083d99a2","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2025-10-10T21:43:50Z","title_canon_sha256":"ac8e2cde2e0a5981f7ccbd503bdf9f1b6993a0e2a55da1770bad22efc1ee93e2"},"schema_version":"1.0","source":{"id":"2510.09885","kind":"arxiv","version":6}},"canonical_sha256":"8807577326141dfa95566b8b6dc2392562d3520316a17d2ff70b9bd2e7b769c0","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"8807577326141dfa95566b8b6dc2392562d3520316a17d2ff70b9bd2e7b769c0","first_computed_at":"2026-06-11T01:09:19.890707Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-11T01:09:19.890707Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"hU8HAYl1NuvCI+9h9xt0DJfSoG32aiprIuqGs1jjzwS1vZEADF+Uiss/C9/lDiIv6UtZWsVtkA4bzp8QVm04Dg==","signature_status":"signed_v1","signed_at":"2026-06-11T01:09:19.891532Z","signed_message":"canonical_sha256_bytes"},"source_id":"2510.09885","source_kind":"arxiv","source_version":6}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:6a33c8a9f42032c6e46117dfd0781a81a60e06935d9ecad48100cfc4a327efdc","sha256:07d2c8b4f8cbdf6e644887ff8fccb4c5765aec1ee7176acca6c6e5457ddc5dac"],"state_sha256":"53aa862b76e2cb0b3184bceb77e9ea862be268f55096c46c809f0304f20b5ca6"}