{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:32AGNVESK7C6YI24VJUC4G546R","short_pith_number":"pith:32AGNVES","canonical_record":{"source":{"id":"2502.08524","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-02-12T16:00:11Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"7fe977c8f17f7b8ba023c59cff72900b5b66bd6e5d01f452accf094fb260ed69","abstract_canon_sha256":"0b47c7623be47e9f9c7e19780f79d3b11ebbbcfdd9776dbeeff9a3fe66a0317d"},"schema_version":"1.0"},"canonical_sha256":"de8066d49257c5ec235caa682e1bbcf446f38f4239e1848df90b5c9d030327c8","source":{"kind":"arxiv","id":"2502.08524","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.08524","created_at":"2026-07-05T10:13:24Z"},{"alias_kind":"arxiv_version","alias_value":"2502.08524v1","created_at":"2026-07-05T10:13:24Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.08524","created_at":"2026-07-05T10:13:24Z"},{"alias_kind":"pith_short_12","alias_value":"32AGNVESK7C6","created_at":"2026-07-05T10:13:24Z"},{"alias_kind":"pith_short_16","alias_value":"32AGNVESK7C6YI24","created_at":"2026-07-05T10:13:24Z"},{"alias_kind":"pith_short_8","alias_value":"32AGNVES","created_at":"2026-07-05T10:13:24Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:32AGNVESK7C6YI24VJUC4G546R","target":"record","payload":{"canonical_record":{"source":{"id":"2502.08524","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-02-12T16:00:11Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"7fe977c8f17f7b8ba023c59cff72900b5b66bd6e5d01f452accf094fb260ed69","abstract_canon_sha256":"0b47c7623be47e9f9c7e19780f79d3b11ebbbcfdd9776dbeeff9a3fe66a0317d"},"schema_version":"1.0"},"canonical_sha256":"de8066d49257c5ec235caa682e1bbcf446f38f4239e1848df90b5c9d030327c8","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:13:24.359162Z","signature_b64":"BHOF7WE1zE1SkpxskD1hRlTHSx1UkwgnFiPMX8nS21p8XOBohPduKz42d6Q7+NSaO+OA9YDQ38+2OjoRp0frDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"de8066d49257c5ec235caa682e1bbcf446f38f4239e1848df90b5c9d030327c8","last_reissued_at":"2026-07-05T10:13:24.358671Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:13:24.358671Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2502.08524","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":"2026-07-05T10:13:24Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"DAidxy/jQJMvla6WZCllFZJgkmtchbZQiRPk93VaJ0gTRM8zBWWfUXP8RZVdQqeJ7Heu22zZiHiXAbHlu1C9Dg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-04T17:10:29.105242Z"},"content_sha256":"cb8b0aed4bee1ccb8def5bab4c05b8f649f173d1123ef9346da866279986fcf4","schema_version":"1.0","event_id":"sha256:cb8b0aed4bee1ccb8def5bab4c05b8f649f173d1123ef9346da866279986fcf4"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:32AGNVESK7C6YI24VJUC4G546R","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"LLM Pretraining with Continuous Concepts","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Andrew Cohen, Ilia Kulikov, Jack Lanchantin, Jane Yu, Janice Lan, Jason Weston, Jihoon Tack, Shibo Hao, Xian Li, Yuandong Tian","submitted_at":"2025-02-12T16:00:11Z","abstract_excerpt":"Next token prediction has been the standard training objective used in large language model pretraining. Representations are learned as a result of optimizing for token-level perplexity. We propose Continuous Concept Mixing (CoCoMix), a novel pretraining framework that combines discrete next token prediction with continuous concepts. Specifically, CoCoMix predicts continuous concepts learned from a pretrained sparse autoencoder and mixes them into the model's hidden state by interleaving with token hidden representations. Through experiments on multiple benchmarks, including language modeling "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.08524","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/2502.08524/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:13:24Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"vdm09QTao+/ASkDyo7PTjAKWZsPqIUMu78isBphPHsUTF4H7GCiqj93sJkASq4TvH5w2v+ERrpQemDAQGukTCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-04T17:10:29.105804Z"},"content_sha256":"0be44e136df9a2f08972191391012fd958e0735c67a0efb88490b4d0434ff3f4","schema_version":"1.0","event_id":"sha256:0be44e136df9a2f08972191391012fd958e0735c67a0efb88490b4d0434ff3f4"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/32AGNVESK7C6YI24VJUC4G546R/bundle.json","state_url":"https://pith.science/pith/32AGNVESK7C6YI24VJUC4G546R/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/32AGNVESK7C6YI24VJUC4G546R/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":"2026-08-04T17:10:29Z","links":{"resolver":"https://pith.science/pith/32AGNVESK7C6YI24VJUC4G546R","bundle":"https://pith.science/pith/32AGNVESK7C6YI24VJUC4G546R/bundle.json","state":"https://pith.science/pith/32AGNVESK7C6YI24VJUC4G546R/state.json","well_known_bundle":"https://pith.science/.well-known/pith/32AGNVESK7C6YI24VJUC4G546R/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:32AGNVESK7C6YI24VJUC4G546R","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":"0b47c7623be47e9f9c7e19780f79d3b11ebbbcfdd9776dbeeff9a3fe66a0317d","cross_cats_sorted":["cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-02-12T16:00:11Z","title_canon_sha256":"7fe977c8f17f7b8ba023c59cff72900b5b66bd6e5d01f452accf094fb260ed69"},"schema_version":"1.0","source":{"id":"2502.08524","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.08524","created_at":"2026-07-05T10:13:24Z"},{"alias_kind":"arxiv_version","alias_value":"2502.08524v1","created_at":"2026-07-05T10:13:24Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.08524","created_at":"2026-07-05T10:13:24Z"},{"alias_kind":"pith_short_12","alias_value":"32AGNVESK7C6","created_at":"2026-07-05T10:13:24Z"},{"alias_kind":"pith_short_16","alias_value":"32AGNVESK7C6YI24","created_at":"2026-07-05T10:13:24Z"},{"alias_kind":"pith_short_8","alias_value":"32AGNVES","created_at":"2026-07-05T10:13:24Z"}],"graph_snapshots":[{"event_id":"sha256:0be44e136df9a2f08972191391012fd958e0735c67a0efb88490b4d0434ff3f4","target":"graph","created_at":"2026-07-05T10:13:24Z","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"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2502.08524/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Next token prediction has been the standard training objective used in large language model pretraining. Representations are learned as a result of optimizing for token-level perplexity. We propose Continuous Concept Mixing (CoCoMix), a novel pretraining framework that combines discrete next token prediction with continuous concepts. Specifically, CoCoMix predicts continuous concepts learned from a pretrained sparse autoencoder and mixes them into the model's hidden state by interleaving with token hidden representations. Through experiments on multiple benchmarks, including language modeling ","authors_text":"Andrew Cohen, Ilia Kulikov, Jack Lanchantin, Jane Yu, Janice Lan, Jason Weston, Jihoon Tack, Shibo Hao, Xian Li, Yuandong Tian","cross_cats":["cs.CL"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-02-12T16:00:11Z","title":"LLM Pretraining with Continuous Concepts"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.08524","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:cb8b0aed4bee1ccb8def5bab4c05b8f649f173d1123ef9346da866279986fcf4","target":"record","created_at":"2026-07-05T10:13:24Z","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":"0b47c7623be47e9f9c7e19780f79d3b11ebbbcfdd9776dbeeff9a3fe66a0317d","cross_cats_sorted":["cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-02-12T16:00:11Z","title_canon_sha256":"7fe977c8f17f7b8ba023c59cff72900b5b66bd6e5d01f452accf094fb260ed69"},"schema_version":"1.0","source":{"id":"2502.08524","kind":"arxiv","version":1}},"canonical_sha256":"de8066d49257c5ec235caa682e1bbcf446f38f4239e1848df90b5c9d030327c8","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"de8066d49257c5ec235caa682e1bbcf446f38f4239e1848df90b5c9d030327c8","first_computed_at":"2026-07-05T10:13:24.358671Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:13:24.358671Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"BHOF7WE1zE1SkpxskD1hRlTHSx1UkwgnFiPMX8nS21p8XOBohPduKz42d6Q7+NSaO+OA9YDQ38+2OjoRp0frDA==","signature_status":"signed_v1","signed_at":"2026-07-05T10:13:24.359162Z","signed_message":"canonical_sha256_bytes"},"source_id":"2502.08524","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:cb8b0aed4bee1ccb8def5bab4c05b8f649f173d1123ef9346da866279986fcf4","sha256:0be44e136df9a2f08972191391012fd958e0735c67a0efb88490b4d0434ff3f4"],"state_sha256":"de9c4a9d7d9a314c6cdabe65d26ef053e441ae8f9c44f29c6a90f9a448c73111"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"nsyPAXS+uvf+MPvu3rmHLBaUCpo0ZU2GBXV8HnKxt4Ni70DWwH38J5ZX1xGm6XroyD/eZLDExfNM6ZecQMgbCw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-04T17:10:29.109319Z","bundle_sha256":"8ca8043e1090249cebf0fef447604618c970168d678fe62547bfe0e5cd969d4d"}}