{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:IH7UWDP3NIIJAFCU2AQ4QTGOT6","short_pith_number":"pith:IH7UWDP3","schema_version":"1.0","canonical_sha256":"41ff4b0dfb6a10901454d021c84cce9f818fd3cac553a32106053f35fdf7251b","source":{"kind":"arxiv","id":"2407.10490","version":4},"attestation_state":"computed","paper":{"title":"Learning Dynamics of LLM Finetuning","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Danica J. Sutherland, Yi Ren","submitted_at":"2024-07-15T07:30:28Z","abstract_excerpt":"Learning dynamics, which describes how the learning of specific training examples influences the model's predictions on other examples, gives us a powerful tool for understanding the behavior of deep learning systems. We study the learning dynamics of large language models during different types of finetuning, by analyzing the step-wise decomposition of how influence accumulates among different potential responses. Our framework allows a uniform interpretation of many interesting observations about the training of popular algorithms for both instruction tuning and preference tuning. In particu"},"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":"2407.10490","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2024-07-15T07:30:28Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"da39c21d5ce1c6793857b505b4062c44643812a02e81023fe89c96ff67d8e497","abstract_canon_sha256":"fc639cd066b4676b61fd5fcde1d9e3961886f5b25f8684f9d517e09dca8ff2e3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:29:00.372182Z","signature_b64":"F0BRqvSJ+mWTx+ALXhcW7UJ82DhZPV6xy7/uF5sZbHzJXDRwL7k/BLm+JCgwAHppE2r+lRXxUnW/l2EmQhUCDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"41ff4b0dfb6a10901454d021c84cce9f818fd3cac553a32106053f35fdf7251b","last_reissued_at":"2026-07-05T11:29:00.371715Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:29:00.371715Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning Dynamics of LLM Finetuning","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Danica J. Sutherland, Yi Ren","submitted_at":"2024-07-15T07:30:28Z","abstract_excerpt":"Learning dynamics, which describes how the learning of specific training examples influences the model's predictions on other examples, gives us a powerful tool for understanding the behavior of deep learning systems. We study the learning dynamics of large language models during different types of finetuning, by analyzing the step-wise decomposition of how influence accumulates among different potential responses. Our framework allows a uniform interpretation of many interesting observations about the training of popular algorithms for both instruction tuning and preference tuning. In particu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.10490","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/2407.10490/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":"2407.10490","created_at":"2026-07-05T11:29:00.371766+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.10490v4","created_at":"2026-07-05T11:29:00.371766+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.10490","created_at":"2026-07-05T11:29:00.371766+00:00"},{"alias_kind":"pith_short_12","alias_value":"IH7UWDP3NIIJ","created_at":"2026-07-05T11:29:00.371766+00:00"},{"alias_kind":"pith_short_16","alias_value":"IH7UWDP3NIIJAFCU","created_at":"2026-07-05T11:29:00.371766+00:00"},{"alias_kind":"pith_short_8","alias_value":"IH7UWDP3","created_at":"2026-07-05T11:29:00.371766+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":8,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.01558","citing_title":"Attention-guided Fine-tuning of Multimodal Large Language Models Improves Chain-of-Thought Reasoning","ref_index":60,"is_internal_anchor":false},{"citing_arxiv_id":"2512.00778","citing_title":"What Is Preference Optimization Doing, and Why?","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2602.23964","citing_title":"RAD-DPO: Robust Adaptive Denoising Direct Preference Optimization for Generative Retrieval in E-commerce","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08666","citing_title":"The Cancellation Hypothesis in Critic-Free RL: From Outcome Rewards to Token Credits","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08378","citing_title":"Reinforcement Learning for Scalable and Trustworthy Intelligent Systems","ref_index":88,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07353","citing_title":"Confidence-Aware Alignment Makes Reasoning LLMs More Reliable","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16995","citing_title":"SPS: Steering Probability Squeezing for Better Exploration in Reinforcement Learning for Large Language Models","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2605.02626","citing_title":"Gradient-Gated DPO: Stabilizing Preference Optimization in Language Models","ref_index":11,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IH7UWDP3NIIJAFCU2AQ4QTGOT6","json":"https://pith.science/pith/IH7UWDP3NIIJAFCU2AQ4QTGOT6.json","graph_json":"https://pith.science/api/pith-number/IH7UWDP3NIIJAFCU2AQ4QTGOT6/graph.json","events_json":"https://pith.science/api/pith-number/IH7UWDP3NIIJAFCU2AQ4QTGOT6/events.json","paper":"https://pith.science/paper/IH7UWDP3"},"agent_actions":{"view_html":"https://pith.science/pith/IH7UWDP3NIIJAFCU2AQ4QTGOT6","download_json":"https://pith.science/pith/IH7UWDP3NIIJAFCU2AQ4QTGOT6.json","view_paper":"https://pith.science/paper/IH7UWDP3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.10490&json=true","fetch_graph":"https://pith.science/api/pith-number/IH7UWDP3NIIJAFCU2AQ4QTGOT6/graph.json","fetch_events":"https://pith.science/api/pith-number/IH7UWDP3NIIJAFCU2AQ4QTGOT6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IH7UWDP3NIIJAFCU2AQ4QTGOT6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IH7UWDP3NIIJAFCU2AQ4QTGOT6/action/storage_attestation","attest_author":"https://pith.science/pith/IH7UWDP3NIIJAFCU2AQ4QTGOT6/action/author_attestation","sign_citation":"https://pith.science/pith/IH7UWDP3NIIJAFCU2AQ4QTGOT6/action/citation_signature","submit_replication":"https://pith.science/pith/IH7UWDP3NIIJAFCU2AQ4QTGOT6/action/replication_record"}},"created_at":"2026-07-05T11:29:00.371766+00:00","updated_at":"2026-07-05T11:29:00.371766+00:00"}