{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:PAUB754NUSFKBUN3N3PAARDVZC","short_pith_number":"pith:PAUB754N","canonical_record":{"source":{"id":"2605.28705","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-27T16:31:51Z","cross_cats_sorted":[],"title_canon_sha256":"7404d3d94fda61788b308b678a28d1c95d6730131fdf3a4937f133403b8f0cdc","abstract_canon_sha256":"be142cda924364350b8f398ba150556a41cc8ad855dea4b4cf95391a0031ec07"},"schema_version":"1.0"},"canonical_sha256":"78281ff78da48aa0d1bb6ede004475c89db2f655f0b2b66ea5884cf6d61f83d2","source":{"kind":"arxiv","id":"2605.28705","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.28705","created_at":"2026-05-28T02:05:00Z"},{"alias_kind":"arxiv_version","alias_value":"2605.28705v1","created_at":"2026-05-28T02:05:00Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.28705","created_at":"2026-05-28T02:05:00Z"},{"alias_kind":"pith_short_12","alias_value":"PAUB754NUSFK","created_at":"2026-05-28T02:05:00Z"},{"alias_kind":"pith_short_16","alias_value":"PAUB754NUSFKBUN3","created_at":"2026-05-28T02:05:00Z"},{"alias_kind":"pith_short_8","alias_value":"PAUB754N","created_at":"2026-05-28T02:05:00Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:PAUB754NUSFKBUN3N3PAARDVZC","target":"record","payload":{"canonical_record":{"source":{"id":"2605.28705","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-27T16:31:51Z","cross_cats_sorted":[],"title_canon_sha256":"7404d3d94fda61788b308b678a28d1c95d6730131fdf3a4937f133403b8f0cdc","abstract_canon_sha256":"be142cda924364350b8f398ba150556a41cc8ad855dea4b4cf95391a0031ec07"},"schema_version":"1.0"},"canonical_sha256":"78281ff78da48aa0d1bb6ede004475c89db2f655f0b2b66ea5884cf6d61f83d2","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-28T02:05:00.575245Z","signature_b64":"GbrEudMz/uHVT04pTLsDvmz4ko1iD7OpztxXkGsJugxEnJU6SSlCCE7AldK0WsrQgiPFbW4P9K/naZ9YzDm4CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"78281ff78da48aa0d1bb6ede004475c89db2f655f0b2b66ea5884cf6d61f83d2","last_reissued_at":"2026-05-28T02:05:00.574818Z","signature_status":"signed_v1","first_computed_at":"2026-05-28T02:05:00.574818Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.28705","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":"2026-05-28T02:05:00Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Ddow/5X+/2b/SbqfqOvKtDizEMiHZ5I3Ah8je80kIMiIm41bRS3oLIylaF4c8wE88pBMbfp8Q4GtXnMsE4voDw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-02T08:04:35.230305Z"},"content_sha256":"acda6ccf9a154cb9b1a11ebd41704cd0b3d1adf453c8d8bd940c8d8897b17114","schema_version":"1.0","event_id":"sha256:acda6ccf9a154cb9b1a11ebd41704cd0b3d1adf453c8d8bd940c8d8897b17114"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:PAUB754NUSFKBUN3N3PAARDVZC","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Understanding Generalization and Forgetting in In-Context Continual Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Guangyu Li, Lijie Hu, Meng Ding","submitted_at":"2026-05-27T16:31:51Z","abstract_excerpt":"In-context learning (ICL) derives its power from enabling Large Language Models to adapt to new tasks via prompt-based reasoning alone, entirely bypassing the need for parameter updates. Existing theories primarily study ICL in single-task settings, while real-world prompts often contain sequences of heterogeneous tasks, leaving a gap in understanding whether Large Language Models implicitly perform continual learning during inference. To bridge this gap, we propose the first theoretical framework for in-context continual learning, modeling how a pretrained Transformer processes multiple seque"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.28705","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/2605.28705/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":"2026-05-28T02:05:00Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"gz3np4Lo1Y+ivjoktwiJALci+vPwypbN0F3Z+9mJcnGjOWACgii8rGDbbSvhYObOkV1iEqgGEZwBbu44JxYlBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-02T08:04:35.230696Z"},"content_sha256":"c0141c54442e376609d790a5ec2bafa5410fc4e5517e1372b5bad5b4aa501212","schema_version":"1.0","event_id":"sha256:c0141c54442e376609d790a5ec2bafa5410fc4e5517e1372b5bad5b4aa501212"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/PAUB754NUSFKBUN3N3PAARDVZC/bundle.json","state_url":"https://pith.science/pith/PAUB754NUSFKBUN3N3PAARDVZC/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/PAUB754NUSFKBUN3N3PAARDVZC/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":"2026-06-02T08:04:35Z","links":{"resolver":"https://pith.science/pith/PAUB754NUSFKBUN3N3PAARDVZC","bundle":"https://pith.science/pith/PAUB754NUSFKBUN3N3PAARDVZC/bundle.json","state":"https://pith.science/pith/PAUB754NUSFKBUN3N3PAARDVZC/state.json","well_known_bundle":"https://pith.science/.well-known/pith/PAUB754NUSFKBUN3N3PAARDVZC/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:PAUB754NUSFKBUN3N3PAARDVZC","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":"be142cda924364350b8f398ba150556a41cc8ad855dea4b4cf95391a0031ec07","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-27T16:31:51Z","title_canon_sha256":"7404d3d94fda61788b308b678a28d1c95d6730131fdf3a4937f133403b8f0cdc"},"schema_version":"1.0","source":{"id":"2605.28705","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.28705","created_at":"2026-05-28T02:05:00Z"},{"alias_kind":"arxiv_version","alias_value":"2605.28705v1","created_at":"2026-05-28T02:05:00Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.28705","created_at":"2026-05-28T02:05:00Z"},{"alias_kind":"pith_short_12","alias_value":"PAUB754NUSFK","created_at":"2026-05-28T02:05:00Z"},{"alias_kind":"pith_short_16","alias_value":"PAUB754NUSFKBUN3","created_at":"2026-05-28T02:05:00Z"},{"alias_kind":"pith_short_8","alias_value":"PAUB754N","created_at":"2026-05-28T02:05:00Z"}],"graph_snapshots":[{"event_id":"sha256:c0141c54442e376609d790a5ec2bafa5410fc4e5517e1372b5bad5b4aa501212","target":"graph","created_at":"2026-05-28T02:05:00Z","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/2605.28705/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"In-context learning (ICL) derives its power from enabling Large Language Models to adapt to new tasks via prompt-based reasoning alone, entirely bypassing the need for parameter updates. Existing theories primarily study ICL in single-task settings, while real-world prompts often contain sequences of heterogeneous tasks, leaving a gap in understanding whether Large Language Models implicitly perform continual learning during inference. To bridge this gap, we propose the first theoretical framework for in-context continual learning, modeling how a pretrained Transformer processes multiple seque","authors_text":"Guangyu Li, Lijie Hu, Meng Ding","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-27T16:31:51Z","title":"Understanding Generalization and Forgetting in In-Context Continual Learning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.28705","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:acda6ccf9a154cb9b1a11ebd41704cd0b3d1adf453c8d8bd940c8d8897b17114","target":"record","created_at":"2026-05-28T02:05:00Z","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":"be142cda924364350b8f398ba150556a41cc8ad855dea4b4cf95391a0031ec07","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-27T16:31:51Z","title_canon_sha256":"7404d3d94fda61788b308b678a28d1c95d6730131fdf3a4937f133403b8f0cdc"},"schema_version":"1.0","source":{"id":"2605.28705","kind":"arxiv","version":1}},"canonical_sha256":"78281ff78da48aa0d1bb6ede004475c89db2f655f0b2b66ea5884cf6d61f83d2","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"78281ff78da48aa0d1bb6ede004475c89db2f655f0b2b66ea5884cf6d61f83d2","first_computed_at":"2026-05-28T02:05:00.574818Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-28T02:05:00.574818Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"GbrEudMz/uHVT04pTLsDvmz4ko1iD7OpztxXkGsJugxEnJU6SSlCCE7AldK0WsrQgiPFbW4P9K/naZ9YzDm4CQ==","signature_status":"signed_v1","signed_at":"2026-05-28T02:05:00.575245Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.28705","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:acda6ccf9a154cb9b1a11ebd41704cd0b3d1adf453c8d8bd940c8d8897b17114","sha256:c0141c54442e376609d790a5ec2bafa5410fc4e5517e1372b5bad5b4aa501212"],"state_sha256":"e4c8570483fb7fc0549783563190039d9085b3285a69379a28a4660eacd61652"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"9DlxcqiO3lFSPtSjaPIdkyZ1G4BfSg6v21CZzOTZfyLDaXqiu4QxZNceeZ9w4li99FInbzt6xZjRar+aIODyDQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-02T08:04:35.232766Z","bundle_sha256":"85c026e9da3ba893b2b67bb6e6e4476ed4161ecd8f7fee053f677441666ec232"}}