{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:HERP3XJWGKBLXFCKH43YA2DIKA","short_pith_number":"pith:HERP3XJW","schema_version":"1.0","canonical_sha256":"3922fddd363282bb944a3f378068685024fd0176a9695a7d7ce33906d4179aa9","source":{"kind":"arxiv","id":"2503.04548","version":1},"attestation_state":"computed","paper":{"title":"An Empirical Study on Eliciting and Improving R1-like Reasoning Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Beichen Zhang, Daixuan Cheng, Jie Chen, Jinhao Jiang, Ji-Rong Wen, Lei Fang, Wayne Xin Zhao, Xu Miao, Yang Lu, Yingqian Min, Zheng Liu, Zhipeng Chen, Zhongyuan Wang","submitted_at":"2025-03-06T15:34:27Z","abstract_excerpt":"In this report, we present the third technical report on the development of slow-thinking models as part of the STILL project. As the technical pathway becomes clearer, scaling RL training has become a central technique for implementing such reasoning models. We systematically experiment with and document the effects of various factors influencing RL training, conducting experiments on both base models and fine-tuned models. Specifically, we demonstrate that our RL training approach consistently improves the Qwen2.5-32B base models, enhancing both response length and test accuracy. Furthermore"},"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":"2503.04548","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-03-06T15:34:27Z","cross_cats_sorted":[],"title_canon_sha256":"e5c2b9604850faeb9c42285cd0ab880eed757c3357e4a688f26cf7d11e89edb7","abstract_canon_sha256":"c0232b05a38c377f789c7f1c4a9ac10c96960fb073be7058042932814cd3634f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:25:41.807476Z","signature_b64":"gpxCYJpoTVYdN7BkZ48rrRRWVtXpycPFaX4SgZZRz+jnLZVUjaezlHMGpzlnU0jioE3YDNT1rlQt29/pksQgBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3922fddd363282bb944a3f378068685024fd0176a9695a7d7ce33906d4179aa9","last_reissued_at":"2026-07-05T10:25:41.806921Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:25:41.806921Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"An Empirical Study on Eliciting and Improving R1-like Reasoning Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Beichen Zhang, Daixuan Cheng, Jie Chen, Jinhao Jiang, Ji-Rong Wen, Lei Fang, Wayne Xin Zhao, Xu Miao, Yang Lu, Yingqian Min, Zheng Liu, Zhipeng Chen, Zhongyuan Wang","submitted_at":"2025-03-06T15:34:27Z","abstract_excerpt":"In this report, we present the third technical report on the development of slow-thinking models as part of the STILL project. As the technical pathway becomes clearer, scaling RL training has become a central technique for implementing such reasoning models. We systematically experiment with and document the effects of various factors influencing RL training, conducting experiments on both base models and fine-tuned models. Specifically, we demonstrate that our RL training approach consistently improves the Qwen2.5-32B base models, enhancing both response length and test accuracy. Furthermore"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.04548","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/2503.04548/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":"2503.04548","created_at":"2026-07-05T10:25:41.807005+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.04548v1","created_at":"2026-07-05T10:25:41.807005+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.04548","created_at":"2026-07-05T10:25:41.807005+00:00"},{"alias_kind":"pith_short_12","alias_value":"HERP3XJWGKBL","created_at":"2026-07-05T10:25:41.807005+00:00"},{"alias_kind":"pith_short_16","alias_value":"HERP3XJWGKBLXFCK","created_at":"2026-07-05T10:25:41.807005+00:00"},{"alias_kind":"pith_short_8","alias_value":"HERP3XJW","created_at":"2026-07-05T10:25:41.807005+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":15,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.10522","citing_title":"GUI-AC: Enhancing Continual Learning in GUI Agents","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01249","citing_title":"Trust Region On-Policy Distillation","ref_index":289,"is_internal_anchor":false},{"citing_arxiv_id":"2605.26971","citing_title":"RLVR Datasets and Where to Find Them: Tracing Data Lineage for Better Training Data","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2503.14476","citing_title":"DAPO: An Open-Source LLM Reinforcement Learning System at Scale","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2504.13818","citing_title":"Not All Rollouts are Useful: Down-Sampling Rollouts in LLM Reinforcement Learning","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18864","citing_title":"SAGE: Shaping Anchors for Guided Exploration in RLVR of LLMs","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2503.17352","citing_title":"OpenVLThinker: Complex Vision-Language Reasoning via Iterative SFT-RL Cycles","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2509.21882","citing_title":"Position: The Hidden Costs and Measurement Gaps of Reinforcement Learning with Verifiable Rewards","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2510.05837","citing_title":"EEPO: Exploration-Enhanced Policy Optimization via Sample-Then-Forget","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2509.08827","citing_title":"A Survey of Reinforcement Learning for Large Reasoning Models","ref_index":82,"is_internal_anchor":false},{"citing_arxiv_id":"2504.21776","citing_title":"WebThinker: Empowering Large Reasoning Models with Deep Research Capability","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12518","citing_title":"TimelineReasoner: Advancing Timeline Summarization with Large Reasoning Models","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2504.11536","citing_title":"ReTool: Reinforcement Learning for Strategic Tool Use in LLMs","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09931","citing_title":"PruneTIR: Inference-Time Tool Call Pruning for Effective yet Efficient Tool-Integrated Reasoning","ref_index":40,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08817","citing_title":"How You Begin is How You Reason: Driving Exploration in RLVR via Prefix-Tuned Priors","ref_index":6,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HERP3XJWGKBLXFCKH43YA2DIKA","json":"https://pith.science/pith/HERP3XJWGKBLXFCKH43YA2DIKA.json","graph_json":"https://pith.science/api/pith-number/HERP3XJWGKBLXFCKH43YA2DIKA/graph.json","events_json":"https://pith.science/api/pith-number/HERP3XJWGKBLXFCKH43YA2DIKA/events.json","paper":"https://pith.science/paper/HERP3XJW"},"agent_actions":{"view_html":"https://pith.science/pith/HERP3XJWGKBLXFCKH43YA2DIKA","download_json":"https://pith.science/pith/HERP3XJWGKBLXFCKH43YA2DIKA.json","view_paper":"https://pith.science/paper/HERP3XJW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.04548&json=true","fetch_graph":"https://pith.science/api/pith-number/HERP3XJWGKBLXFCKH43YA2DIKA/graph.json","fetch_events":"https://pith.science/api/pith-number/HERP3XJWGKBLXFCKH43YA2DIKA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HERP3XJWGKBLXFCKH43YA2DIKA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HERP3XJWGKBLXFCKH43YA2DIKA/action/storage_attestation","attest_author":"https://pith.science/pith/HERP3XJWGKBLXFCKH43YA2DIKA/action/author_attestation","sign_citation":"https://pith.science/pith/HERP3XJWGKBLXFCKH43YA2DIKA/action/citation_signature","submit_replication":"https://pith.science/pith/HERP3XJWGKBLXFCKH43YA2DIKA/action/replication_record"}},"created_at":"2026-07-05T10:25:41.807005+00:00","updated_at":"2026-07-05T10:25:41.807005+00:00"}