{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:RVWQJCQYLIODWCD47I5JL5X3PD","short_pith_number":"pith:RVWQJCQY","schema_version":"1.0","canonical_sha256":"8d6d048a185a1c3b087cfa3a95f6fb78c658658c4e98a5e7f5f2bb8ed238adf6","source":{"kind":"arxiv","id":"2602.22642","version":2},"attestation_state":"computed","paper":{"title":"Compress the Easy, Explore the Hard: Difficulty-Aware Entropy Regularization for Efficient LLM Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Jun Fang, Naiqiang Tan, Qin-Wen Luo, Rui Liu, Sheng-Jun Huang, Sheng Ren, Xiang Chen","submitted_at":"2026-02-26T05:47:30Z","abstract_excerpt":"Chain-of-Thought (CoT) has substantially empowered Large Language Models (LLMs) to tackle complex reasoning tasks, yet the verbose nature of explicit reasoning steps incurs prohibitive inference latency and computational costs, limiting real-world deployment. While existing compression methods - ranging from self-training to Reinforcement Learning (RL) with length constraints - attempt to mitigate this, they often sacrifice reasoning capability for brevity. We identify a critical failure mode in these approaches: explicitly optimizing for shorter trajectories triggers rapid entropy collapse, w"},"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":"2602.22642","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-26T05:47:30Z","cross_cats_sorted":[],"title_canon_sha256":"b8be7c6db345b1cb0903f68c5dda983fc4159145bd5baa12ee2e7272f3e87a68","abstract_canon_sha256":"3b90e231ebca01208aeb10119401c6225014384708e4587cf27e81536732faeb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-23T00:11:49.020786Z","signature_b64":"Wnjr1iStIeFdk7R7ZwPQO/o2XUQsg+wGhh7n8iyL410v6vsojkUUkr/kcD820fEj4RUK6wpWP7jRnYRdj/yZDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8d6d048a185a1c3b087cfa3a95f6fb78c658658c4e98a5e7f5f2bb8ed238adf6","last_reissued_at":"2026-06-23T00:11:49.020270Z","signature_status":"signed_v1","first_computed_at":"2026-06-23T00:11:49.020270Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Compress the Easy, Explore the Hard: Difficulty-Aware Entropy Regularization for Efficient LLM Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Jun Fang, Naiqiang Tan, Qin-Wen Luo, Rui Liu, Sheng-Jun Huang, Sheng Ren, Xiang Chen","submitted_at":"2026-02-26T05:47:30Z","abstract_excerpt":"Chain-of-Thought (CoT) has substantially empowered Large Language Models (LLMs) to tackle complex reasoning tasks, yet the verbose nature of explicit reasoning steps incurs prohibitive inference latency and computational costs, limiting real-world deployment. While existing compression methods - ranging from self-training to Reinforcement Learning (RL) with length constraints - attempt to mitigate this, they often sacrifice reasoning capability for brevity. We identify a critical failure mode in these approaches: explicitly optimizing for shorter trajectories triggers rapid entropy collapse, w"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2602.22642","kind":"arxiv","version":2},"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/2602.22642/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":"2602.22642","created_at":"2026-06-23T00:11:49.020349+00:00"},{"alias_kind":"arxiv_version","alias_value":"2602.22642v2","created_at":"2026-06-23T00:11:49.020349+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2602.22642","created_at":"2026-06-23T00:11:49.020349+00:00"},{"alias_kind":"pith_short_12","alias_value":"RVWQJCQYLIOD","created_at":"2026-06-23T00:11:49.020349+00:00"},{"alias_kind":"pith_short_16","alias_value":"RVWQJCQYLIODWCD4","created_at":"2026-06-23T00:11:49.020349+00:00"},{"alias_kind":"pith_short_8","alias_value":"RVWQJCQY","created_at":"2026-06-23T00:11:49.020349+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.07316","citing_title":"Implicit Compression Regularization: Concise Reasoning via Internal Shorter Distributions in RL Post-Training","ref_index":22,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RVWQJCQYLIODWCD47I5JL5X3PD","json":"https://pith.science/pith/RVWQJCQYLIODWCD47I5JL5X3PD.json","graph_json":"https://pith.science/api/pith-number/RVWQJCQYLIODWCD47I5JL5X3PD/graph.json","events_json":"https://pith.science/api/pith-number/RVWQJCQYLIODWCD47I5JL5X3PD/events.json","paper":"https://pith.science/paper/RVWQJCQY"},"agent_actions":{"view_html":"https://pith.science/pith/RVWQJCQYLIODWCD47I5JL5X3PD","download_json":"https://pith.science/pith/RVWQJCQYLIODWCD47I5JL5X3PD.json","view_paper":"https://pith.science/paper/RVWQJCQY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2602.22642&json=true","fetch_graph":"https://pith.science/api/pith-number/RVWQJCQYLIODWCD47I5JL5X3PD/graph.json","fetch_events":"https://pith.science/api/pith-number/RVWQJCQYLIODWCD47I5JL5X3PD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RVWQJCQYLIODWCD47I5JL5X3PD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RVWQJCQYLIODWCD47I5JL5X3PD/action/storage_attestation","attest_author":"https://pith.science/pith/RVWQJCQYLIODWCD47I5JL5X3PD/action/author_attestation","sign_citation":"https://pith.science/pith/RVWQJCQYLIODWCD47I5JL5X3PD/action/citation_signature","submit_replication":"https://pith.science/pith/RVWQJCQYLIODWCD47I5JL5X3PD/action/replication_record"}},"created_at":"2026-06-23T00:11:49.020349+00:00","updated_at":"2026-06-23T00:11:49.020349+00:00"}