{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:KBVDFN5QIFUWCCKRLN36ZROBRD","short_pith_number":"pith:KBVDFN5Q","canonical_record":{"source":{"id":"2605.15053","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T16:46:26Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"ea116ec5ee880a739d3230409ddf937660666bf91e6438614a1993dea2da79b3","abstract_canon_sha256":"2b54313f538f8fd7d43050c491eb0c45c47a06d1d483fdfbd79d8c7d093f7bad"},"schema_version":"1.0"},"canonical_sha256":"506a32b7b041696109515b77ecc5c188eb0de3ad691bd476f49a63d236cca1f9","source":{"kind":"arxiv","id":"2605.15053","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.15053","created_at":"2026-05-17T23:38:54Z"},{"alias_kind":"arxiv_version","alias_value":"2605.15053v1","created_at":"2026-05-17T23:38:54Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.15053","created_at":"2026-05-17T23:38:54Z"},{"alias_kind":"pith_short_12","alias_value":"KBVDFN5QIFUW","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"KBVDFN5QIFUWCCKR","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"KBVDFN5Q","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:KBVDFN5QIFUWCCKRLN36ZROBRD","target":"record","payload":{"canonical_record":{"source":{"id":"2605.15053","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T16:46:26Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"ea116ec5ee880a739d3230409ddf937660666bf91e6438614a1993dea2da79b3","abstract_canon_sha256":"2b54313f538f8fd7d43050c491eb0c45c47a06d1d483fdfbd79d8c7d093f7bad"},"schema_version":"1.0"},"canonical_sha256":"506a32b7b041696109515b77ecc5c188eb0de3ad691bd476f49a63d236cca1f9","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:54.369554Z","signature_b64":"rXUcZ9l/dXO80vuGvKobVT+aT4exGeCmTr1138vlIGNGgT6q6KKwy8BnYLHhkygulJXYRzaiRHq+wa4pTF9dDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"506a32b7b041696109515b77ecc5c188eb0de3ad691bd476f49a63d236cca1f9","last_reissued_at":"2026-05-17T23:38:54.368863Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:54.368863Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.15053","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:54Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"hPznNPi9lD0CscXun/WDK/lrCZEbo/Jsv2Jd9dhAvAhVf9/RgQ68oLYwcj6dN0ZJAH5Ao/ahHNoCMerzWRk0Dg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-21T03:33:59.665787Z"},"content_sha256":"361a52e66cab5ba82c58ea5295403c52197f318cfc1d071cc6f26fc04e63f5ee","schema_version":"1.0","event_id":"sha256:361a52e66cab5ba82c58ea5295403c52197f318cfc1d071cc6f26fc04e63f5ee"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:KBVDFN5QIFUWCCKRLN36ZROBRD","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"TFGN: Task-Free, Replay-Free Continual Pre-Training Without Catastrophic Forgetting at LLM Scale","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Anurup Ganguli","submitted_at":"2026-05-14T16:46:26Z","abstract_excerpt":"Continually pre-training a large language model on heterogeneous text domains, without replay or task labels, has remained an unsolved architectural problem at LLM scale. Existing methods rely on replay buffers, task identifiers, regularization penalties that scale poorly, or sentence-classification-scale evaluation.\n  We introduce TFGN, an architectural overlay for transformer language models that produces input-conditioned, parameter-efficient updates while leaving the rest of the transformer unchanged. On six heterogeneous text domains (Prose, Python, Math, Biomedical, Chinese, JavaScript) "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.15053","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":""},"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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:54Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"fuRWkIYn3z1LPaiNwuL05agpWnjarMiX6tnlymzevJW6rWkVv5IW28o+dJLupBF26AhyA0lWG7qJbsMOWfPIBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-21T03:33:59.666133Z"},"content_sha256":"9f50ef933f7bc2581c6b193fe3ff70360262e69ca585270d01d46a2ab11d6e43","schema_version":"1.0","event_id":"sha256:9f50ef933f7bc2581c6b193fe3ff70360262e69ca585270d01d46a2ab11d6e43"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/KBVDFN5QIFUWCCKRLN36ZROBRD/bundle.json","state_url":"https://pith.science/pith/KBVDFN5QIFUWCCKRLN36ZROBRD/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/KBVDFN5QIFUWCCKRLN36ZROBRD/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-05-21T03:33:59Z","links":{"resolver":"https://pith.science/pith/KBVDFN5QIFUWCCKRLN36ZROBRD","bundle":"https://pith.science/pith/KBVDFN5QIFUWCCKRLN36ZROBRD/bundle.json","state":"https://pith.science/pith/KBVDFN5QIFUWCCKRLN36ZROBRD/state.json","well_known_bundle":"https://pith.science/.well-known/pith/KBVDFN5QIFUWCCKRLN36ZROBRD/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:KBVDFN5QIFUWCCKRLN36ZROBRD","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":"2b54313f538f8fd7d43050c491eb0c45c47a06d1d483fdfbd79d8c7d093f7bad","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T16:46:26Z","title_canon_sha256":"ea116ec5ee880a739d3230409ddf937660666bf91e6438614a1993dea2da79b3"},"schema_version":"1.0","source":{"id":"2605.15053","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.15053","created_at":"2026-05-17T23:38:54Z"},{"alias_kind":"arxiv_version","alias_value":"2605.15053v1","created_at":"2026-05-17T23:38:54Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.15053","created_at":"2026-05-17T23:38:54Z"},{"alias_kind":"pith_short_12","alias_value":"KBVDFN5QIFUW","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"KBVDFN5QIFUWCCKR","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"KBVDFN5Q","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:9f50ef933f7bc2581c6b193fe3ff70360262e69ca585270d01d46a2ab11d6e43","target":"graph","created_at":"2026-05-17T23:38:54Z","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":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"Continually pre-training a large language model on heterogeneous text domains, without replay or task labels, has remained an unsolved architectural problem at LLM scale. Existing methods rely on replay buffers, task identifiers, regularization penalties that scale poorly, or sentence-classification-scale evaluation.\n  We introduce TFGN, an architectural overlay for transformer language models that produces input-conditioned, parameter-efficient updates while leaving the rest of the transformer unchanged. On six heterogeneous text domains (Prose, Python, Math, Biomedical, Chinese, JavaScript) ","authors_text":"Anurup Ganguli","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T16:46:26Z","title":"TFGN: Task-Free, Replay-Free Continual Pre-Training Without Catastrophic Forgetting at LLM Scale"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.15053","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:361a52e66cab5ba82c58ea5295403c52197f318cfc1d071cc6f26fc04e63f5ee","target":"record","created_at":"2026-05-17T23:38:54Z","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":"2b54313f538f8fd7d43050c491eb0c45c47a06d1d483fdfbd79d8c7d093f7bad","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-14T16:46:26Z","title_canon_sha256":"ea116ec5ee880a739d3230409ddf937660666bf91e6438614a1993dea2da79b3"},"schema_version":"1.0","source":{"id":"2605.15053","kind":"arxiv","version":1}},"canonical_sha256":"506a32b7b041696109515b77ecc5c188eb0de3ad691bd476f49a63d236cca1f9","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"506a32b7b041696109515b77ecc5c188eb0de3ad691bd476f49a63d236cca1f9","first_computed_at":"2026-05-17T23:38:54.368863Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:54.368863Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"rXUcZ9l/dXO80vuGvKobVT+aT4exGeCmTr1138vlIGNGgT6q6KKwy8BnYLHhkygulJXYRzaiRHq+wa4pTF9dDg==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:54.369554Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.15053","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:361a52e66cab5ba82c58ea5295403c52197f318cfc1d071cc6f26fc04e63f5ee","sha256:9f50ef933f7bc2581c6b193fe3ff70360262e69ca585270d01d46a2ab11d6e43"],"state_sha256":"f5841bd1442affbbb75159b8deaee796c5e80f16e12613b2ddabdae3c5f78f21"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"e3On9ZZrQWErIfM+Zl2PIrh0iQBho7HY6l4BRA4jZHxWF/mWw9znwiJpg7d18HMm4zEuT5PV1oszqF3YUipACg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-21T03:33:59.668121Z","bundle_sha256":"89865f3565cb687d8ce935713dcfc6979a7baff002201727f8a9a5ae2904c77b"}}