{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:RTQGP7BP4LMGQEVOYJLLX5NVJT","short_pith_number":"pith:RTQGP7BP","canonical_record":{"source":{"id":"2601.05488","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-01-09T02:44:37Z","cross_cats_sorted":[],"title_canon_sha256":"c178304caa6aaa1b902a468ea4ee41cc68c2669cb31e2b34459160bab673c3cb","abstract_canon_sha256":"e5a186dec8385e48b46cebc9616f85230615985ba34d607dc23fbbfe0d0e4a56"},"schema_version":"1.0"},"canonical_sha256":"8ce067fc2fe2d86812aec256bbf5b54ceb75bb201f921243cb2addeb7842f823","source":{"kind":"arxiv","id":"2601.05488","version":4},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2601.05488","created_at":"2026-06-23T02:13:19Z"},{"alias_kind":"arxiv_version","alias_value":"2601.05488v4","created_at":"2026-06-23T02:13:19Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2601.05488","created_at":"2026-06-23T02:13:19Z"},{"alias_kind":"pith_short_12","alias_value":"RTQGP7BP4LMG","created_at":"2026-06-23T02:13:19Z"},{"alias_kind":"pith_short_16","alias_value":"RTQGP7BP4LMGQEVO","created_at":"2026-06-23T02:13:19Z"},{"alias_kind":"pith_short_8","alias_value":"RTQGP7BP","created_at":"2026-06-23T02:13:19Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:RTQGP7BP4LMGQEVOYJLLX5NVJT","target":"record","payload":{"canonical_record":{"source":{"id":"2601.05488","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-01-09T02:44:37Z","cross_cats_sorted":[],"title_canon_sha256":"c178304caa6aaa1b902a468ea4ee41cc68c2669cb31e2b34459160bab673c3cb","abstract_canon_sha256":"e5a186dec8385e48b46cebc9616f85230615985ba34d607dc23fbbfe0d0e4a56"},"schema_version":"1.0"},"canonical_sha256":"8ce067fc2fe2d86812aec256bbf5b54ceb75bb201f921243cb2addeb7842f823","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-23T02:13:19.896643Z","signature_b64":"UfgWRVT80O/mZchGVuw4OL+x33hVowqaW9lWWS6wLQsoDkBejU0utmSQ6nb/HJSV5qiCIjmUyHiWzrq37EfRDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8ce067fc2fe2d86812aec256bbf5b54ceb75bb201f921243cb2addeb7842f823","last_reissued_at":"2026-06-23T02:13:19.896189Z","signature_status":"signed_v1","first_computed_at":"2026-06-23T02:13:19.896189Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2601.05488","source_version":4,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-23T02:13:19Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"jY5SXeAbpbRpUsBu/cxyDPdv6o6GF1PuqVNbQ0PuQfmmJqKkaVLRQuEt9KZqAmq914FAy3rA/K+sqp7PumTBDw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-27T16:34:11.529785Z"},"content_sha256":"8681eb1597ab36c323d9a9e6d255735c42e7c66e53e8d5eedc02c8d81d3ace27","schema_version":"1.0","event_id":"sha256:8681eb1597ab36c323d9a9e6d255735c42e7c66e53e8d5eedc02c8d81d3ace27"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:RTQGP7BP4LMGQEVOYJLLX5NVJT","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"MemBuilder: Reinforcing LLMs for Long-Term Memory Construction via Attributed Dense Rewards","license":"http://creativecommons.org/licenses/by/4.0/","headline":"MemBuilder trains a 4B model with dense rewards to outperform closed-source LLMs on long-term dialogue memory.","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Fuming Lai, Shaobing Lian, Yanghui Rao, Zhiyu Shen, Ziming Wu","submitted_at":"2026-01-09T02:44:37Z","abstract_excerpt":"Maintaining consistency in long-term dialogues remains a fundamental challenge for LLMs, as standard retrieval mechanisms often fail to capture the temporal evolution of historical states. While memory-augmented frameworks offer a structured alternative, current systems rely on static prompting of closed-source models or suffer from ineffective training paradigms with sparse rewards. We introduce MemBuilder, a reinforcement learning framework that trains models to orchestrate multi-dimensional memory construction with attributed dense rewards. MemBuilder addresses two key challenges: (1) Spars"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Experimental results show that MemBuilder enables a 4B-parameter model to outperform state-of-the-art closed-source baselines, exhibiting strong generalization across long-term dialogue benchmarks.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That synthetic session-level question generation produces unbiased, representative dense rewards that accurately measure and improve real-world memory construction quality without introducing artifacts.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"MemBuilder trains 4B-parameter models with attributed dense rewards to outperform closed-source baselines on long-term dialogue memory tasks.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"MemBuilder trains a 4B model with dense rewards to outperform closed-source LLMs on long-term dialogue memory.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"ccfbef3e30dbb23547d333e01a49c691f8be3498c592a377ffa851f87fa0fcca"},"source":{"id":"2601.05488","kind":"arxiv","version":4},"verdict":{"id":"0838669c-3178-4531-b8bf-c0848244cc2f","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T16:53:01.591957Z","strongest_claim":"Experimental results show that MemBuilder enables a 4B-parameter model to outperform state-of-the-art closed-source baselines, exhibiting strong generalization across long-term dialogue benchmarks.","one_line_summary":"MemBuilder trains 4B-parameter models with attributed dense rewards to outperform closed-source baselines on long-term dialogue memory tasks.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That synthetic session-level question generation produces unbiased, representative dense rewards that accurately measure and improve real-world memory construction quality without introducing artifacts.","pith_extraction_headline":"MemBuilder trains a 4B model with dense rewards to outperform closed-source LLMs on long-term dialogue memory."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2601.05488/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":2,"snapshot_sha256":"05d0f612b60166ebd6c38efa8142ad949ea10fca898c3dcf954e947f08f7fc1b"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"0838669c-3178-4531-b8bf-c0848244cc2f"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-23T02:13:19Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"wD3yXnlfMPIYoI84zk6l53QJGK+Gfjg1uFXmBEsazWZFdZuh0QHU5VCRJlWTIaki04WaTtBfXZ623SFyKdA1CQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-27T16:34:11.530228Z"},"content_sha256":"c5d20a8e137c25bcd4f3a1896471435eb2dd6ef30647374a65eb81dff54282e2","schema_version":"1.0","event_id":"sha256:c5d20a8e137c25bcd4f3a1896471435eb2dd6ef30647374a65eb81dff54282e2"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/RTQGP7BP4LMGQEVOYJLLX5NVJT/bundle.json","state_url":"https://pith.science/pith/RTQGP7BP4LMGQEVOYJLLX5NVJT/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/RTQGP7BP4LMGQEVOYJLLX5NVJT/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":"2026-06-27T16:34:11Z","links":{"resolver":"https://pith.science/pith/RTQGP7BP4LMGQEVOYJLLX5NVJT","bundle":"https://pith.science/pith/RTQGP7BP4LMGQEVOYJLLX5NVJT/bundle.json","state":"https://pith.science/pith/RTQGP7BP4LMGQEVOYJLLX5NVJT/state.json","well_known_bundle":"https://pith.science/.well-known/pith/RTQGP7BP4LMGQEVOYJLLX5NVJT/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:RTQGP7BP4LMGQEVOYJLLX5NVJT","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":"e5a186dec8385e48b46cebc9616f85230615985ba34d607dc23fbbfe0d0e4a56","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-01-09T02:44:37Z","title_canon_sha256":"c178304caa6aaa1b902a468ea4ee41cc68c2669cb31e2b34459160bab673c3cb"},"schema_version":"1.0","source":{"id":"2601.05488","kind":"arxiv","version":4}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2601.05488","created_at":"2026-06-23T02:13:19Z"},{"alias_kind":"arxiv_version","alias_value":"2601.05488v4","created_at":"2026-06-23T02:13:19Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2601.05488","created_at":"2026-06-23T02:13:19Z"},{"alias_kind":"pith_short_12","alias_value":"RTQGP7BP4LMG","created_at":"2026-06-23T02:13:19Z"},{"alias_kind":"pith_short_16","alias_value":"RTQGP7BP4LMGQEVO","created_at":"2026-06-23T02:13:19Z"},{"alias_kind":"pith_short_8","alias_value":"RTQGP7BP","created_at":"2026-06-23T02:13:19Z"}],"graph_snapshots":[{"event_id":"sha256:c5d20a8e137c25bcd4f3a1896471435eb2dd6ef30647374a65eb81dff54282e2","target":"graph","created_at":"2026-06-23T02:13:19Z","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":"Experimental results show that MemBuilder enables a 4B-parameter model to outperform state-of-the-art closed-source baselines, exhibiting strong generalization across long-term dialogue benchmarks."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That synthetic session-level question generation produces unbiased, representative dense rewards that accurately measure and improve real-world memory construction quality without introducing artifacts."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"MemBuilder trains 4B-parameter models with attributed dense rewards to outperform closed-source baselines on long-term dialogue memory tasks."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"MemBuilder trains a 4B model with dense rewards to outperform closed-source LLMs on long-term dialogue memory."}],"snapshot_sha256":"ccfbef3e30dbb23547d333e01a49c691f8be3498c592a377ffa851f87fa0fcca"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"05d0f612b60166ebd6c38efa8142ad949ea10fca898c3dcf954e947f08f7fc1b"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2601.05488/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Maintaining consistency in long-term dialogues remains a fundamental challenge for LLMs, as standard retrieval mechanisms often fail to capture the temporal evolution of historical states. While memory-augmented frameworks offer a structured alternative, current systems rely on static prompting of closed-source models or suffer from ineffective training paradigms with sparse rewards. We introduce MemBuilder, a reinforcement learning framework that trains models to orchestrate multi-dimensional memory construction with attributed dense rewards. MemBuilder addresses two key challenges: (1) Spars","authors_text":"Fuming Lai, Shaobing Lian, Yanghui Rao, Zhiyu Shen, Ziming Wu","cross_cats":[],"headline":"MemBuilder trains a 4B model with dense rewards to outperform closed-source LLMs on long-term dialogue memory.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-01-09T02:44:37Z","title":"MemBuilder: Reinforcing LLMs for Long-Term Memory Construction via Attributed Dense Rewards"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2601.05488","kind":"arxiv","version":4},"verdict":{"created_at":"2026-05-16T16:53:01.591957Z","id":"0838669c-3178-4531-b8bf-c0848244cc2f","model_set":{"reader":"grok-4.3"},"one_line_summary":"MemBuilder trains 4B-parameter models with attributed dense rewards to outperform closed-source baselines on long-term dialogue memory tasks.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"MemBuilder trains a 4B model with dense rewards to outperform closed-source LLMs on long-term dialogue memory.","strongest_claim":"Experimental results show that MemBuilder enables a 4B-parameter model to outperform state-of-the-art closed-source baselines, exhibiting strong generalization across long-term dialogue benchmarks.","weakest_assumption":"That synthetic session-level question generation produces unbiased, representative dense rewards that accurately measure and improve real-world memory construction quality without introducing artifacts."}},"verdict_id":"0838669c-3178-4531-b8bf-c0848244cc2f"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:8681eb1597ab36c323d9a9e6d255735c42e7c66e53e8d5eedc02c8d81d3ace27","target":"record","created_at":"2026-06-23T02:13:19Z","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":"e5a186dec8385e48b46cebc9616f85230615985ba34d607dc23fbbfe0d0e4a56","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-01-09T02:44:37Z","title_canon_sha256":"c178304caa6aaa1b902a468ea4ee41cc68c2669cb31e2b34459160bab673c3cb"},"schema_version":"1.0","source":{"id":"2601.05488","kind":"arxiv","version":4}},"canonical_sha256":"8ce067fc2fe2d86812aec256bbf5b54ceb75bb201f921243cb2addeb7842f823","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"8ce067fc2fe2d86812aec256bbf5b54ceb75bb201f921243cb2addeb7842f823","first_computed_at":"2026-06-23T02:13:19.896189Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-23T02:13:19.896189Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"UfgWRVT80O/mZchGVuw4OL+x33hVowqaW9lWWS6wLQsoDkBejU0utmSQ6nb/HJSV5qiCIjmUyHiWzrq37EfRDQ==","signature_status":"signed_v1","signed_at":"2026-06-23T02:13:19.896643Z","signed_message":"canonical_sha256_bytes"},"source_id":"2601.05488","source_kind":"arxiv","source_version":4}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:8681eb1597ab36c323d9a9e6d255735c42e7c66e53e8d5eedc02c8d81d3ace27","sha256:c5d20a8e137c25bcd4f3a1896471435eb2dd6ef30647374a65eb81dff54282e2"],"state_sha256":"920e1731664bdf1b08ff527ca2515a1eb94d0fe4f80ea3da35a9439b24814ea5"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"AgaCWLrETrFouPmkkU23zNeAYjEppWM2iwZIrzjlwSH7ZctXPn5LUkOaPnahsDAFi1RyFWHOkD2DEQI3EOGzAQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-27T16:34:11.532349Z","bundle_sha256":"1e7a568d7054ccb34d5215f21c4d9f5858b96ebabf877bfc6289bd4eaf8db088"}}