{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:VIPPQC5TP22IXB77XMII7BNZ6J","short_pith_number":"pith:VIPPQC5T","schema_version":"1.0","canonical_sha256":"aa1ef80bb37eb48b87ffbb108f85b9f26d5816c48658efe80dd9402bb2b84343","source":{"kind":"arxiv","id":"2508.03682","version":4},"attestation_state":"computed","paper":{"title":"Self-Questioning Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Deepak Pathak, Hao Liu, Katerina Fragkiadaki, Lili Chen, Mihir Prabhudesai","submitted_at":"2025-08-05T17:51:33Z","abstract_excerpt":"Can large language models improve without external data -- by generating their own questions and answers? We hypothesize that a pre-trained language model can improve its reasoning skills given only a single prompt specifying the topic (e.g., algebra word problems) and asking the model to generate its own questions. To do this, we propose Self-Questioning Language Models (SQLM): an asymmetric self-play framework where a proposer is given the topic and generates a question for a solver, who tries to answer it. Both the proposer and solver are trained via reinforcement learning. The proposer rec"},"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":"2508.03682","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-08-05T17:51:33Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"0a16729182233b755fe1421fb7ce861453bf30410a9ebfc2a680274f095ca0bd","abstract_canon_sha256":"4cdbbdf81a62437cdddf48381cf0b3041048e2f593198c3a120a99d09bec8ef3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:08:05.931888Z","signature_b64":"LN4OtY+gig6OgpF8QHMtfRQL4Uj8EYtF7dD6gE1Em98iYmW4XFd2Spo5t+/1ffiffznis8G+GaTSnx1W3XQgAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"aa1ef80bb37eb48b87ffbb108f85b9f26d5816c48658efe80dd9402bb2b84343","last_reissued_at":"2026-07-05T12:08:05.931157Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:08:05.931157Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Self-Questioning Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Deepak Pathak, Hao Liu, Katerina Fragkiadaki, Lili Chen, Mihir Prabhudesai","submitted_at":"2025-08-05T17:51:33Z","abstract_excerpt":"Can large language models improve without external data -- by generating their own questions and answers? We hypothesize that a pre-trained language model can improve its reasoning skills given only a single prompt specifying the topic (e.g., algebra word problems) and asking the model to generate its own questions. To do this, we propose Self-Questioning Language Models (SQLM): an asymmetric self-play framework where a proposer is given the topic and generates a question for a solver, who tries to answer it. Both the proposer and solver are trained via reinforcement learning. The proposer rec"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.03682","kind":"arxiv","version":4},"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/2508.03682/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":"2508.03682","created_at":"2026-07-05T12:08:05.931241+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.03682v4","created_at":"2026-07-05T12:08:05.931241+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.03682","created_at":"2026-07-05T12:08:05.931241+00:00"},{"alias_kind":"pith_short_12","alias_value":"VIPPQC5TP22I","created_at":"2026-07-05T12:08:05.931241+00:00"},{"alias_kind":"pith_short_16","alias_value":"VIPPQC5TP22IXB77","created_at":"2026-07-05T12:08:05.931241+00:00"},{"alias_kind":"pith_short_8","alias_value":"VIPPQC5T","created_at":"2026-07-05T12:08:05.931241+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":13,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.17682","citing_title":"From Trainee to Trainer: LLM-Designed Training Environment for RL with Multi-Agent Reasoning","ref_index":103,"is_internal_anchor":false},{"citing_arxiv_id":"2606.11719","citing_title":"Ouroboros-Spatial: Closing the Data-Model Loop for Spatial Reasoning","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01249","citing_title":"Trust Region On-Policy Distillation","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01248","citing_title":"$S^3$-R1: Learning to Retrieve and Answer Step-by-Step with Synthetic Data","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2512.18552","citing_title":"Toward Training Superintelligent Software Agents through Self-Play SWE-RL","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20189","citing_title":"SOLAR: A Self-Optimizing Open-Ended Autonomous Agent for Lifelong Learning and Continual Adaptation","ref_index":62,"is_internal_anchor":false},{"citing_arxiv_id":"2509.08827","citing_title":"A Survey of Reinforcement Learning for Large Reasoning Models","ref_index":63,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08516","citing_title":"OracleTSC: Oracle-Informed Reward Hurdle and Uncertainty Regularization for Traffic Signal Control","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01248","citing_title":"$S^3$-R1: Learning to Retrieve and Answer Step-by-Step with Synthetic Data","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2604.20209","citing_title":"Scaling Self-Play with Self-Guidance","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2604.07774","citing_title":"RoboAgent: Chaining Basic Capabilities for Embodied Task Planning","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2604.07864","citing_title":"ZeroCoder: Can LLMs Improve Code Generation Without Ground-Truth Supervision?","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2604.14054","citing_title":"$\\pi$-Play: Multi-Agent Self-Play via Privileged Self-Distillation without External Data","ref_index":3,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VIPPQC5TP22IXB77XMII7BNZ6J","json":"https://pith.science/pith/VIPPQC5TP22IXB77XMII7BNZ6J.json","graph_json":"https://pith.science/api/pith-number/VIPPQC5TP22IXB77XMII7BNZ6J/graph.json","events_json":"https://pith.science/api/pith-number/VIPPQC5TP22IXB77XMII7BNZ6J/events.json","paper":"https://pith.science/paper/VIPPQC5T"},"agent_actions":{"view_html":"https://pith.science/pith/VIPPQC5TP22IXB77XMII7BNZ6J","download_json":"https://pith.science/pith/VIPPQC5TP22IXB77XMII7BNZ6J.json","view_paper":"https://pith.science/paper/VIPPQC5T","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.03682&json=true","fetch_graph":"https://pith.science/api/pith-number/VIPPQC5TP22IXB77XMII7BNZ6J/graph.json","fetch_events":"https://pith.science/api/pith-number/VIPPQC5TP22IXB77XMII7BNZ6J/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VIPPQC5TP22IXB77XMII7BNZ6J/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VIPPQC5TP22IXB77XMII7BNZ6J/action/storage_attestation","attest_author":"https://pith.science/pith/VIPPQC5TP22IXB77XMII7BNZ6J/action/author_attestation","sign_citation":"https://pith.science/pith/VIPPQC5TP22IXB77XMII7BNZ6J/action/citation_signature","submit_replication":"https://pith.science/pith/VIPPQC5TP22IXB77XMII7BNZ6J/action/replication_record"}},"created_at":"2026-07-05T12:08:05.931241+00:00","updated_at":"2026-07-05T12:08:05.931241+00:00"}