{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:3AKEZE3TAPIIHZYC5RBY7ZDHZN","short_pith_number":"pith:3AKEZE3T","schema_version":"1.0","canonical_sha256":"d8144c937303d083e702ec438fe467cb66da96ccf76f3bf79fd0a307195bf4b8","source":{"kind":"arxiv","id":"2507.02259","version":1},"attestation_state":"computed","paper":{"title":"MemAgent: Reshaping Long-Context LLM with Multi-Conv RL-based Memory Agent","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"MemAgent lets LLMs handle millions of tokens by segmenting input and overwriting memory after RL training on 32K texts.","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Hao Zhou, Hongli Yu, Jiangjie Chen, Jiangtao Feng, Jingjing Liu, Mingxuan Wang, Qiying Yu, Tinghong Chen, Weinan Dai, Wei-Ying Ma, Ya-Qin Zhang","submitted_at":"2025-07-03T03:11:50Z","abstract_excerpt":"Despite improvements by length extrapolation, efficient attention and memory modules, handling infinitely long documents with linear complexity without performance degradation during extrapolation remains the ultimate challenge in long-text processing. We directly optimize for long-text tasks in an end-to-end fashion and introduce a novel agent workflow, MemAgent, which reads text in segments and updates the memory using an overwrite strategy. We extend the DAPO algorithm to facilitate training via independent-context multi-conversation generation. MemAgent has demonstrated superb long-context"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":true,"formal_links_present":true},"canonical_record":{"source":{"id":"2507.02259","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-07-03T03:11:50Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"1c5f2eb4f7d03e2225f06250d7bc510eab14efdecea0d98d161a35291af95019","abstract_canon_sha256":"909f67bdf57361e1e01dbde35b145d65f5a3e5e6d9083acc5b7b887a65a2b28d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:52.699547Z","signature_b64":"XWNAIYQaeb5zNw5Lm6Ln3O2q/+WK3vJh4XaAF4oRqu6xlGwnpkVn5JmzUAQn4m2hXsTUFDePrqO7DRuE9752DQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d8144c937303d083e702ec438fe467cb66da96ccf76f3bf79fd0a307195bf4b8","last_reissued_at":"2026-05-17T23:38:52.698861Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:52.698861Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MemAgent: Reshaping Long-Context LLM with Multi-Conv RL-based Memory Agent","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"MemAgent lets LLMs handle millions of tokens by segmenting input and overwriting memory after RL training on 32K texts.","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Hao Zhou, Hongli Yu, Jiangjie Chen, Jiangtao Feng, Jingjing Liu, Mingxuan Wang, Qiying Yu, Tinghong Chen, Weinan Dai, Wei-Ying Ma, Ya-Qin Zhang","submitted_at":"2025-07-03T03:11:50Z","abstract_excerpt":"Despite improvements by length extrapolation, efficient attention and memory modules, handling infinitely long documents with linear complexity without performance degradation during extrapolation remains the ultimate challenge in long-text processing. We directly optimize for long-text tasks in an end-to-end fashion and introduce a novel agent workflow, MemAgent, which reads text in segments and updates the memory using an overwrite strategy. We extend the DAPO algorithm to facilitate training via independent-context multi-conversation generation. MemAgent has demonstrated superb long-context"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"MemAgent has demonstrated superb long-context capabilities, being able to extrapolate from an 8K context trained on 32K text to a 3.5M QA task with performance loss < 5% and achieves 95%+ in 512K RULER test.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the overwrite memory strategy combined with multi-conversation RL training will continue to prevent performance degradation when scaling far beyond the 32K training length without additional mechanisms or data.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"MemAgent uses multi-conversation RL to train a memory agent that reads text in segments and overwrites memory, extrapolating from 8K training to 3.5M token QA with under 5% loss and 95%+ on 512K RULER.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"MemAgent lets LLMs handle millions of tokens by segmenting input and overwriting memory after RL training on 32K texts.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"37d6df6170b4e6d89a70129ae69baaeb847622fd5e1472e0ab5d8c9f2783130d"},"source":{"id":"2507.02259","kind":"arxiv","version":1},"verdict":{"id":"49cf8c9c-1fb9-4c13-b384-d5862d5733c3","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T11:11:50.392302Z","strongest_claim":"MemAgent has demonstrated superb long-context capabilities, being able to extrapolate from an 8K context trained on 32K text to a 3.5M QA task with performance loss < 5% and achieves 95%+ in 512K RULER test.","one_line_summary":"MemAgent uses multi-conversation RL to train a memory agent that reads text in segments and overwrites memory, extrapolating from 8K training to 3.5M token QA with under 5% loss and 95%+ on 512K RULER.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the overwrite memory strategy combined with multi-conversation RL training will continue to prevent performance degradation when scaling far beyond the 32K training length without additional mechanisms or data.","pith_extraction_headline":"MemAgent lets LLMs handle millions of tokens by segmenting input and overwriting memory after RL training on 32K texts."},"references":{"count":65,"sample":[{"doi":"","year":2024,"title":"RULER: What's the Real Context Size of Your Long-Context Language Models?","work_id":"c0bc4689-3ce8-4e3d-9442-bd74869445bb","ref_index":1,"cited_arxiv_id":"2404.06654","is_internal_anchor":true},{"doi":"","year":2018,"title":"HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering","work_id":"c87d7e5f-b81a-41c8-beca-f0b9d598aae4","ref_index":2,"cited_arxiv_id":"1809.09600","is_internal_anchor":true},{"doi":"","year":2024,"title":"Learning to reason with llms","work_id":"16f7723e-0b48-4125-885f-310de5bf2c28","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Gemini 2.0 flash thinking, 2024","work_id":"ae6b4683-11f7-4205-91f3-a116f2c49f43","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Grok 3 beta — the age of reasoning agents, 2024","work_id":"21963273-aa23-44a9-be08-9bd3fd2bc191","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":65,"snapshot_sha256":"ab37c378c53f205646ae8c52b2abcc0c15c8085d63d5d0d0c88ec0179c9d0268","internal_anchors":33},"formal_canon":{"evidence_count":2,"snapshot_sha256":"57dc516578b067252ffe60da63e7f3913ced605f33dcf1ea6b02242da768c781"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2507.02259","created_at":"2026-05-17T23:38:52.698966+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.02259v1","created_at":"2026-05-17T23:38:52.698966+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.02259","created_at":"2026-05-17T23:38:52.698966+00:00"},{"alias_kind":"pith_short_12","alias_value":"3AKEZE3TAPII","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"3AKEZE3TAPIIHZYC","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"3AKEZE3T","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":37,"internal_anchor_count":37,"sample":[{"citing_arxiv_id":"2605.23590","citing_title":"Co-ReAct: Rubrics as Step-Level Collaborators for ReAct Agents","ref_index":13,"is_internal_anchor":true},{"citing_arxiv_id":"2605.21768","citing_title":"Memory-R2: Fair Credit Assignment for Long-Horizon Memory-Augmented LLM Agents","ref_index":25,"is_internal_anchor":true},{"citing_arxiv_id":"2601.21468","citing_title":"MemOCR: Layout-Aware Visual Memory for Efficient Long-Horizon Reasoning","ref_index":32,"is_internal_anchor":true},{"citing_arxiv_id":"2603.23231","citing_title":"PERMA: Benchmarking Personalized Memory Agents via Event-Driven Preference and Realistic Task Environments","ref_index":78,"is_internal_anchor":true},{"citing_arxiv_id":"2605.20616","citing_title":"Auto-Dreamer: Learning Offline Memory Consolidation for Language Agents","ref_index":39,"is_internal_anchor":true},{"citing_arxiv_id":"2605.21463","citing_title":"Mem-$\\pi$: Adaptive Memory through Learning When and What to Generate","ref_index":55,"is_internal_anchor":true},{"citing_arxiv_id":"2605.18421","citing_title":"EvoMemBench: Benchmarking Agent Memory from a Self-Evolving Perspective","ref_index":40,"is_internal_anchor":true},{"citing_arxiv_id":"2605.19932","citing_title":"PEEK: Context Map as an Orientation Cache for Long-Context LLM Agents","ref_index":48,"is_internal_anchor":true},{"citing_arxiv_id":"2508.03341","citing_title":"What Deserves Memory: Adaptive Memory Distillation for LLM Agents","ref_index":22,"is_internal_anchor":true},{"citing_arxiv_id":"2509.02547","citing_title":"The Landscape of Agentic Reinforcement Learning for LLMs: A Survey","ref_index":137,"is_internal_anchor":true},{"citing_arxiv_id":"2509.13021","citing_title":"xOffense: An Autonomous Multi-Agent Framework for Penetration Testing with Domain-Adapted Large Language Models","ref_index":40,"is_internal_anchor":true},{"citing_arxiv_id":"2511.02805","citing_title":"MemSearcher: Training LLMs to Reason, Search and Manage Memory via End-to-End Reinforcement Learning","ref_index":36,"is_internal_anchor":true},{"citing_arxiv_id":"2507.05257","citing_title":"Evaluating Memory in LLM Agents via Incremental Multi-Turn Interactions","ref_index":46,"is_internal_anchor":true},{"citing_arxiv_id":"2601.09636","citing_title":"PersonalAlign: Hierarchical Implicit Intent Alignment for Personalized GUI Agent with Long-Term User-Centric Records","ref_index":5,"is_internal_anchor":true},{"citing_arxiv_id":"2602.13933","citing_title":"HyMem: Hybrid Memory Architecture with Dynamic Retrieval Scheduling","ref_index":47,"is_internal_anchor":true},{"citing_arxiv_id":"2603.23516","citing_title":"MSA: Memory Sparse Attention for Efficient End-to-End Memory Model Scaling to 100M Tokens","ref_index":49,"is_internal_anchor":true},{"citing_arxiv_id":"2603.06194","citing_title":"MICA: Multi-granularity Intertemporal Credit Assignment for Long-Horizon Emotional Support Dialogue","ref_index":59,"is_internal_anchor":true},{"citing_arxiv_id":"2603.12572","citing_title":"LMEB: Long-horizon Memory Embedding Benchmark","ref_index":38,"is_internal_anchor":true},{"citing_arxiv_id":"2603.15421","citing_title":"CLAG: Adaptive Memory Organization via Agent-Driven Clustering for Small Language Model Agents","ref_index":7,"is_internal_anchor":true},{"citing_arxiv_id":"2511.20857","citing_title":"Evo-Memory: Benchmarking LLM Agent Test-time Learning with Self-Evolving Memory","ref_index":140,"is_internal_anchor":true},{"citing_arxiv_id":"2603.29002","citing_title":"Understand and Accelerate Memory Processing Pipeline for Large Language Model Inference","ref_index":29,"is_internal_anchor":true},{"citing_arxiv_id":"2603.29493","citing_title":"MemFactory: Unified Inference & Training Framework for Agent Memory","ref_index":17,"is_internal_anchor":true},{"citing_arxiv_id":"2604.02721","citing_title":"GrandCode: Achieving Grandmaster Level in Competitive Programming via Agentic Reinforcement Learning","ref_index":37,"is_internal_anchor":true},{"citing_arxiv_id":"2604.03679","citing_title":"LightThinker++: From Reasoning Compression to Memory Management","ref_index":72,"is_internal_anchor":true},{"citing_arxiv_id":"2605.11814","citing_title":"MedMemoryBench: Benchmarking Agent Memory in Personalized Healthcare","ref_index":39,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3AKEZE3TAPIIHZYC5RBY7ZDHZN","json":"https://pith.science/pith/3AKEZE3TAPIIHZYC5RBY7ZDHZN.json","graph_json":"https://pith.science/api/pith-number/3AKEZE3TAPIIHZYC5RBY7ZDHZN/graph.json","events_json":"https://pith.science/api/pith-number/3AKEZE3TAPIIHZYC5RBY7ZDHZN/events.json","paper":"https://pith.science/paper/3AKEZE3T"},"agent_actions":{"view_html":"https://pith.science/pith/3AKEZE3TAPIIHZYC5RBY7ZDHZN","download_json":"https://pith.science/pith/3AKEZE3TAPIIHZYC5RBY7ZDHZN.json","view_paper":"https://pith.science/paper/3AKEZE3T","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.02259&json=true","fetch_graph":"https://pith.science/api/pith-number/3AKEZE3TAPIIHZYC5RBY7ZDHZN/graph.json","fetch_events":"https://pith.science/api/pith-number/3AKEZE3TAPIIHZYC5RBY7ZDHZN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3AKEZE3TAPIIHZYC5RBY7ZDHZN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3AKEZE3TAPIIHZYC5RBY7ZDHZN/action/storage_attestation","attest_author":"https://pith.science/pith/3AKEZE3TAPIIHZYC5RBY7ZDHZN/action/author_attestation","sign_citation":"https://pith.science/pith/3AKEZE3TAPIIHZYC5RBY7ZDHZN/action/citation_signature","submit_replication":"https://pith.science/pith/3AKEZE3TAPIIHZYC5RBY7ZDHZN/action/replication_record"}},"created_at":"2026-05-17T23:38:52.698966+00:00","updated_at":"2026-05-17T23:38:52.698966+00:00"}