{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:UTYSADMEPEU4EFTBHLS7ECW3AC","short_pith_number":"pith:UTYSADME","schema_version":"1.0","canonical_sha256":"a4f1200d847929c216613ae5f20adb00971a17a016f7bfb19ed360ed88cf392e","source":{"kind":"arxiv","id":"2503.22456","version":2},"attestation_state":"computed","paper":{"title":"Entropy-guided sequence weighting for efficient exploration in RL-based LLM fine-tuning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Abdullah Vanlioglu","submitted_at":"2025-03-28T14:07:51Z","abstract_excerpt":"We introduce Entropy-Guided Sequence Weighting (EGSW), a novel approach that enhances the exploration-exploitation tradeoff by dynamically assigning weights to generated outputs based on their advantage and entropy for Reinforcement Learning-based Large Language Model fine-tuning. EGSW integrates entropy regularization with advantage-based weighting to balance policy updates, enabling efficient exploration in high-dimensional state spaces. By employing temperature-scaled softmax weighting over sequences, EGSW prioritizing high-reward, high-uncertainty steps while maintaining training stability"},"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":"2503.22456","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-03-28T14:07:51Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"5bc8b1b8a5cd55f8efc7a261881eb766f2d159b437693ace45451504e30efbf2","abstract_canon_sha256":"dcacdb40cd64a1e5f95d12d761d0842c997fd79dc23a3f01523c7e53da1e5eb5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:41:55.090452Z","signature_b64":"uXl41MHnA6+9DAtT+7gTUVHiI/oI3xumbmtsgBDj/fI7bvb/uhhBtuXQv4683JDktaUcX0/slcT01yj4CHX4Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a4f1200d847929c216613ae5f20adb00971a17a016f7bfb19ed360ed88cf392e","last_reissued_at":"2026-07-05T10:41:55.089971Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:41:55.089971Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Entropy-guided sequence weighting for efficient exploration in RL-based LLM fine-tuning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Abdullah Vanlioglu","submitted_at":"2025-03-28T14:07:51Z","abstract_excerpt":"We introduce Entropy-Guided Sequence Weighting (EGSW), a novel approach that enhances the exploration-exploitation tradeoff by dynamically assigning weights to generated outputs based on their advantage and entropy for Reinforcement Learning-based Large Language Model fine-tuning. EGSW integrates entropy regularization with advantage-based weighting to balance policy updates, enabling efficient exploration in high-dimensional state spaces. By employing temperature-scaled softmax weighting over sequences, EGSW prioritizing high-reward, high-uncertainty steps while maintaining training stability"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.22456","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/2503.22456/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":"2503.22456","created_at":"2026-07-05T10:41:55.090028+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.22456v2","created_at":"2026-07-05T10:41:55.090028+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.22456","created_at":"2026-07-05T10:41:55.090028+00:00"},{"alias_kind":"pith_short_12","alias_value":"UTYSADMEPEU4","created_at":"2026-07-05T10:41:55.090028+00:00"},{"alias_kind":"pith_short_16","alias_value":"UTYSADMEPEU4EFTB","created_at":"2026-07-05T10:41:55.090028+00:00"},{"alias_kind":"pith_short_8","alias_value":"UTYSADME","created_at":"2026-07-05T10:41:55.090028+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.18089","citing_title":"From Reasoning Traces to Reusable Modules: Understanding Compositional Generalization in Language Model Reasoning","ref_index":127,"is_internal_anchor":false},{"citing_arxiv_id":"2606.06840","citing_title":"Characterize Then Distill: Mechanistic Reasoning in Large Output Spaces","ref_index":142,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12112","citing_title":"When Policy Entropy Constraint Fails: Preserving Diversity in Flow-based RLHF via Perceptual Entropy","ref_index":58,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08978","citing_title":"Learning to Explore: Scaling Agentic Reasoning via Exploration-Aware Policy Optimization","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08978","citing_title":"Learning to Explore: Scaling Agentic Reasoning via Exploration-Aware Policy Optimization","ref_index":17,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UTYSADMEPEU4EFTBHLS7ECW3AC","json":"https://pith.science/pith/UTYSADMEPEU4EFTBHLS7ECW3AC.json","graph_json":"https://pith.science/api/pith-number/UTYSADMEPEU4EFTBHLS7ECW3AC/graph.json","events_json":"https://pith.science/api/pith-number/UTYSADMEPEU4EFTBHLS7ECW3AC/events.json","paper":"https://pith.science/paper/UTYSADME"},"agent_actions":{"view_html":"https://pith.science/pith/UTYSADMEPEU4EFTBHLS7ECW3AC","download_json":"https://pith.science/pith/UTYSADMEPEU4EFTBHLS7ECW3AC.json","view_paper":"https://pith.science/paper/UTYSADME","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.22456&json=true","fetch_graph":"https://pith.science/api/pith-number/UTYSADMEPEU4EFTBHLS7ECW3AC/graph.json","fetch_events":"https://pith.science/api/pith-number/UTYSADMEPEU4EFTBHLS7ECW3AC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UTYSADMEPEU4EFTBHLS7ECW3AC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UTYSADMEPEU4EFTBHLS7ECW3AC/action/storage_attestation","attest_author":"https://pith.science/pith/UTYSADMEPEU4EFTBHLS7ECW3AC/action/author_attestation","sign_citation":"https://pith.science/pith/UTYSADMEPEU4EFTBHLS7ECW3AC/action/citation_signature","submit_replication":"https://pith.science/pith/UTYSADMEPEU4EFTBHLS7ECW3AC/action/replication_record"}},"created_at":"2026-07-05T10:41:55.090028+00:00","updated_at":"2026-07-05T10:41:55.090028+00:00"}