{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:AWFB5HYH2YYC2MQN3OEOFLGKHN","short_pith_number":"pith:AWFB5HYH","schema_version":"1.0","canonical_sha256":"058a1e9f07d6302d320ddb88e2acca3b61f74756d535d84cc0dbe7eab32029db","source":{"kind":"arxiv","id":"2504.00829","version":1},"attestation_state":"computed","paper":{"title":"How Difficulty-Aware Staged Reinforcement Learning Enhances LLMs' Reasoning Capabilities: A Preliminary Experimental Study","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Han Zhao, Haotian Wang, Shuaiting Chen, Sitong Zhao, Xiangang Li, Xiaoyu Tian, Yiping Peng, Yunjie Ji","submitted_at":"2025-04-01T14:18:38Z","abstract_excerpt":"Enhancing the reasoning capabilities of Large Language Models (LLMs) with efficiency and scalability remains a fundamental challenge in artificial intelligence research. This paper presents a rigorous experimental investigation into how difficulty-aware staged reinforcement learning (RL) strategies can substantially improve LLM reasoning performance. Through systematic analysis, we demonstrate that strategically selecting training data according to well-defined difficulty levels markedly enhances RL optimization. Moreover, we introduce a staged training methodology, progressively exposing mode"},"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":"2504.00829","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-04-01T14:18:38Z","cross_cats_sorted":[],"title_canon_sha256":"a9aae1db360d183cdd688272d390c6d9c132e90dce2d3ccc12315fc93fb13215","abstract_canon_sha256":"fedde771d39d5a3143962f6b4651b85c0349f15561b28cc4b1e231ea0e5fc55f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:42:51.051216Z","signature_b64":"qQh52DXAJWTzU6igLGlv4RgNYywDZquRqQQeXfpUB+E6njLa3y3xBo1arhQ+HNsoOkY/tlmjxCfzR7Pk4htkDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"058a1e9f07d6302d320ddb88e2acca3b61f74756d535d84cc0dbe7eab32029db","last_reissued_at":"2026-07-05T10:42:51.050650Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:42:51.050650Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"How Difficulty-Aware Staged Reinforcement Learning Enhances LLMs' Reasoning Capabilities: A Preliminary Experimental Study","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Han Zhao, Haotian Wang, Shuaiting Chen, Sitong Zhao, Xiangang Li, Xiaoyu Tian, Yiping Peng, Yunjie Ji","submitted_at":"2025-04-01T14:18:38Z","abstract_excerpt":"Enhancing the reasoning capabilities of Large Language Models (LLMs) with efficiency and scalability remains a fundamental challenge in artificial intelligence research. This paper presents a rigorous experimental investigation into how difficulty-aware staged reinforcement learning (RL) strategies can substantially improve LLM reasoning performance. Through systematic analysis, we demonstrate that strategically selecting training data according to well-defined difficulty levels markedly enhances RL optimization. Moreover, we introduce a staged training methodology, progressively exposing mode"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.00829","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/2504.00829/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":"2504.00829","created_at":"2026-07-05T10:42:51.050711+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.00829v1","created_at":"2026-07-05T10:42:51.050711+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.00829","created_at":"2026-07-05T10:42:51.050711+00:00"},{"alias_kind":"pith_short_12","alias_value":"AWFB5HYH2YYC","created_at":"2026-07-05T10:42:51.050711+00:00"},{"alias_kind":"pith_short_16","alias_value":"AWFB5HYH2YYC2MQN","created_at":"2026-07-05T10:42:51.050711+00:00"},{"alias_kind":"pith_short_8","alias_value":"AWFB5HYH","created_at":"2026-07-05T10:42:51.050711+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.08480","citing_title":"Adaptive Loss Balancing for Noise-Robust GRPO in Generative Recommendation","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2605.22703","citing_title":"Clipping Bottleneck: Stabilizing RLVR via Stochastic Recovery of Near-Boundary Signals","ref_index":13,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AWFB5HYH2YYC2MQN3OEOFLGKHN","json":"https://pith.science/pith/AWFB5HYH2YYC2MQN3OEOFLGKHN.json","graph_json":"https://pith.science/api/pith-number/AWFB5HYH2YYC2MQN3OEOFLGKHN/graph.json","events_json":"https://pith.science/api/pith-number/AWFB5HYH2YYC2MQN3OEOFLGKHN/events.json","paper":"https://pith.science/paper/AWFB5HYH"},"agent_actions":{"view_html":"https://pith.science/pith/AWFB5HYH2YYC2MQN3OEOFLGKHN","download_json":"https://pith.science/pith/AWFB5HYH2YYC2MQN3OEOFLGKHN.json","view_paper":"https://pith.science/paper/AWFB5HYH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.00829&json=true","fetch_graph":"https://pith.science/api/pith-number/AWFB5HYH2YYC2MQN3OEOFLGKHN/graph.json","fetch_events":"https://pith.science/api/pith-number/AWFB5HYH2YYC2MQN3OEOFLGKHN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AWFB5HYH2YYC2MQN3OEOFLGKHN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AWFB5HYH2YYC2MQN3OEOFLGKHN/action/storage_attestation","attest_author":"https://pith.science/pith/AWFB5HYH2YYC2MQN3OEOFLGKHN/action/author_attestation","sign_citation":"https://pith.science/pith/AWFB5HYH2YYC2MQN3OEOFLGKHN/action/citation_signature","submit_replication":"https://pith.science/pith/AWFB5HYH2YYC2MQN3OEOFLGKHN/action/replication_record"}},"created_at":"2026-07-05T10:42:51.050711+00:00","updated_at":"2026-07-05T10:42:51.050711+00:00"}