{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:VFZ3XYXJP2LESCXICLI7LXRC4R","short_pith_number":"pith:VFZ3XYXJ","schema_version":"1.0","canonical_sha256":"a973bbe2e97e96490ae812d1f5de22e46760bfb139f54a9d7a310500ab2cc101","source":{"kind":"arxiv","id":"2504.10449","version":3},"attestation_state":"computed","paper":{"title":"M1: Towards Scalable Test-Time Compute with Mamba Reasoning Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Alexander M. Rush, Daniele Paliotta, Daniel Ritter, Junxiong Wang, Tri Dao, Wen-Ding Li","submitted_at":"2025-04-14T17:38:25Z","abstract_excerpt":"Effective reasoning is crucial to solving complex mathematical problems. Recent large language models (LLMs) have boosted performance by scaling test-time computation through long chain-of-thought reasoning. However, transformer-based models are inherently limited in extending context length due to their quadratic computational complexity and linear memory requirements. In this paper, we introduce a novel hybrid linear RNN reasoning model, M1, built on the Mamba architecture, which allows memory-efficient inference. Our approach leverages a distillation process from existing reasoning models a"},"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":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2504.10449","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-04-14T17:38:25Z","cross_cats_sorted":[],"title_canon_sha256":"7902997265d3568e87d3590608a49193ddf6edda15c8e2e369ffd4271c469bf2","abstract_canon_sha256":"b608ff2979c59138a78312483b37aea29c4a5a20f98ca3a17991d67eb03232ba"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:07:06.052129Z","signature_b64":"5nHu191alELyg3SLgp4cTKwW3xm/BTirt9+DMwUIrUmIcEmIg2aako/13ws9qj1VyAPldbBvtf69zkUiBfFOCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a973bbe2e97e96490ae812d1f5de22e46760bfb139f54a9d7a310500ab2cc101","last_reissued_at":"2026-07-05T12:07:06.051649Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:07:06.051649Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"M1: Towards Scalable Test-Time Compute with Mamba Reasoning Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Alexander M. Rush, Daniele Paliotta, Daniel Ritter, Junxiong Wang, Tri Dao, Wen-Ding Li","submitted_at":"2025-04-14T17:38:25Z","abstract_excerpt":"Effective reasoning is crucial to solving complex mathematical problems. Recent large language models (LLMs) have boosted performance by scaling test-time computation through long chain-of-thought reasoning. However, transformer-based models are inherently limited in extending context length due to their quadratic computational complexity and linear memory requirements. In this paper, we introduce a novel hybrid linear RNN reasoning model, M1, built on the Mamba architecture, which allows memory-efficient inference. Our approach leverages a distillation process from existing reasoning models a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.10449","kind":"arxiv","version":3},"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/2504.10449/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2504.10449","created_at":"2026-07-05T12:07:06.051708+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.10449v3","created_at":"2026-07-05T12:07:06.051708+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.10449","created_at":"2026-07-05T12:07:06.051708+00:00"},{"alias_kind":"pith_short_12","alias_value":"VFZ3XYXJP2LE","created_at":"2026-07-05T12:07:06.051708+00:00"},{"alias_kind":"pith_short_16","alias_value":"VFZ3XYXJP2LESCXI","created_at":"2026-07-05T12:07:06.051708+00:00"},{"alias_kind":"pith_short_8","alias_value":"VFZ3XYXJ","created_at":"2026-07-05T12:07:06.051708+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.11634","citing_title":"Architecture-Aware Reinforcement Learning Makes Sliding-Window Attention Competitive in Math Reasoning","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2603.04759","citing_title":"Stacked from One: Multi-Scale Self-Injection for Context Window Extension","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2604.24715","citing_title":"Long-Context Aware Upcycling: A New Frontier for Hybrid LLM Scaling","ref_index":48,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VFZ3XYXJP2LESCXICLI7LXRC4R","json":"https://pith.science/pith/VFZ3XYXJP2LESCXICLI7LXRC4R.json","graph_json":"https://pith.science/api/pith-number/VFZ3XYXJP2LESCXICLI7LXRC4R/graph.json","events_json":"https://pith.science/api/pith-number/VFZ3XYXJP2LESCXICLI7LXRC4R/events.json","paper":"https://pith.science/paper/VFZ3XYXJ"},"agent_actions":{"view_html":"https://pith.science/pith/VFZ3XYXJP2LESCXICLI7LXRC4R","download_json":"https://pith.science/pith/VFZ3XYXJP2LESCXICLI7LXRC4R.json","view_paper":"https://pith.science/paper/VFZ3XYXJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.10449&json=true","fetch_graph":"https://pith.science/api/pith-number/VFZ3XYXJP2LESCXICLI7LXRC4R/graph.json","fetch_events":"https://pith.science/api/pith-number/VFZ3XYXJP2LESCXICLI7LXRC4R/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VFZ3XYXJP2LESCXICLI7LXRC4R/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VFZ3XYXJP2LESCXICLI7LXRC4R/action/storage_attestation","attest_author":"https://pith.science/pith/VFZ3XYXJP2LESCXICLI7LXRC4R/action/author_attestation","sign_citation":"https://pith.science/pith/VFZ3XYXJP2LESCXICLI7LXRC4R/action/citation_signature","submit_replication":"https://pith.science/pith/VFZ3XYXJP2LESCXICLI7LXRC4R/action/replication_record"}},"created_at":"2026-07-05T12:07:06.051708+00:00","updated_at":"2026-07-05T12:07:06.051708+00:00"}