{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:BRHDIKEQFYVUEYNYHJOFDMYEPQ","short_pith_number":"pith:BRHDIKEQ","canonical_record":{"source":{"id":"2404.11999","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-04-18T08:49:38Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"c8c0399dac725367951e6eb9468b1dd4d912ba6f1d04e21f3e0447c3010abaf4","abstract_canon_sha256":"dd976dfb6b2c6d8b2bbfe68ea9055987125151c152a7642ca04efae43c0996e9"},"schema_version":"1.0"},"canonical_sha256":"0c4e3428902e2b4261b83a5c51b3047c0a94302027400a610ede05b88bd8065f","source":{"kind":"arxiv","id":"2404.11999","version":5},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2404.11999","created_at":"2026-07-05T09:00:59Z"},{"alias_kind":"arxiv_version","alias_value":"2404.11999v5","created_at":"2026-07-05T09:00:59Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.11999","created_at":"2026-07-05T09:00:59Z"},{"alias_kind":"pith_short_12","alias_value":"BRHDIKEQFYVU","created_at":"2026-07-05T09:00:59Z"},{"alias_kind":"pith_short_16","alias_value":"BRHDIKEQFYVUEYNY","created_at":"2026-07-05T09:00:59Z"},{"alias_kind":"pith_short_8","alias_value":"BRHDIKEQ","created_at":"2026-07-05T09:00:59Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:BRHDIKEQFYVUEYNYHJOFDMYEPQ","target":"record","payload":{"canonical_record":{"source":{"id":"2404.11999","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-04-18T08:49:38Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"c8c0399dac725367951e6eb9468b1dd4d912ba6f1d04e21f3e0447c3010abaf4","abstract_canon_sha256":"dd976dfb6b2c6d8b2bbfe68ea9055987125151c152a7642ca04efae43c0996e9"},"schema_version":"1.0"},"canonical_sha256":"0c4e3428902e2b4261b83a5c51b3047c0a94302027400a610ede05b88bd8065f","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:00:59.505339Z","signature_b64":"RcpSo7bSt9r0QXCsT/tnaYBJ55hbnfB/wmhQ+YxgHv3LsWtyY2S0bZxTIrM2a/CLt0MtO395u51y7Xn7Q1QiCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0c4e3428902e2b4261b83a5c51b3047c0a94302027400a610ede05b88bd8065f","last_reissued_at":"2026-07-05T09:00:59.504850Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:00:59.504850Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2404.11999","source_version":5,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T09:00:59Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"w/8UfZLXXeEfz373HW4appyfDIH+uVwqlW/8cmezTJOVr33B6/GX+vlBuioAqIc3jaMi4aXQYv4IFKCYe3L2AQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T01:56:47.298991Z"},"content_sha256":"4d57b172fe77ef48432605250e529d5957cf6dc10c141d15cb2e7d428e804904","schema_version":"1.0","event_id":"sha256:4d57b172fe77ef48432605250e529d5957cf6dc10c141d15cb2e7d428e804904"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:BRHDIKEQFYVUEYNYHJOFDMYEPQ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Token-level Direct Preference Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Guoqing Liu, Haifeng Zhang, Jun Wang, Ning Yang, Weiyu Ma, Yongcheng Zeng","submitted_at":"2024-04-18T08:49:38Z","abstract_excerpt":"Fine-tuning pre-trained Large Language Models (LLMs) is essential to align them with human values and intentions. This process often utilizes methods like pairwise comparisons and KL divergence against a reference LLM, focusing on the evaluation of full answers generated by the models. However, the generation of these responses occurs in a token level, following a sequential, auto-regressive fashion. In this paper, we introduce Token-level Direct Preference Optimization (TDPO), a novel approach to align LLMs with human preferences by optimizing policy at the token level. Unlike previous method"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.11999","kind":"arxiv","version":5},"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/2404.11999/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T09:00:59Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"IE9sYDuVHzxy/8Mqow2P1pbWJv+iwL/XvVTU659AIhc7yIGExhg/JO366RBP2idDZGcZW2F6FXMkt6hA1d/ECw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T01:56:47.299509Z"},"content_sha256":"0bba5faa53ee4d82b8ec51d232b2c1d0f3c665ca765a9837819c31b014bdf7df","schema_version":"1.0","event_id":"sha256:0bba5faa53ee4d82b8ec51d232b2c1d0f3c665ca765a9837819c31b014bdf7df"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/BRHDIKEQFYVUEYNYHJOFDMYEPQ/bundle.json","state_url":"https://pith.science/pith/BRHDIKEQFYVUEYNYHJOFDMYEPQ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/BRHDIKEQFYVUEYNYHJOFDMYEPQ/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-06T01:56:47Z","links":{"resolver":"https://pith.science/pith/BRHDIKEQFYVUEYNYHJOFDMYEPQ","bundle":"https://pith.science/pith/BRHDIKEQFYVUEYNYHJOFDMYEPQ/bundle.json","state":"https://pith.science/pith/BRHDIKEQFYVUEYNYHJOFDMYEPQ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/BRHDIKEQFYVUEYNYHJOFDMYEPQ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:BRHDIKEQFYVUEYNYHJOFDMYEPQ","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"dd976dfb6b2c6d8b2bbfe68ea9055987125151c152a7642ca04efae43c0996e9","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-04-18T08:49:38Z","title_canon_sha256":"c8c0399dac725367951e6eb9468b1dd4d912ba6f1d04e21f3e0447c3010abaf4"},"schema_version":"1.0","source":{"id":"2404.11999","kind":"arxiv","version":5}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2404.11999","created_at":"2026-07-05T09:00:59Z"},{"alias_kind":"arxiv_version","alias_value":"2404.11999v5","created_at":"2026-07-05T09:00:59Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.11999","created_at":"2026-07-05T09:00:59Z"},{"alias_kind":"pith_short_12","alias_value":"BRHDIKEQFYVU","created_at":"2026-07-05T09:00:59Z"},{"alias_kind":"pith_short_16","alias_value":"BRHDIKEQFYVUEYNY","created_at":"2026-07-05T09:00:59Z"},{"alias_kind":"pith_short_8","alias_value":"BRHDIKEQ","created_at":"2026-07-05T09:00:59Z"}],"graph_snapshots":[{"event_id":"sha256:0bba5faa53ee4d82b8ec51d232b2c1d0f3c665ca765a9837819c31b014bdf7df","target":"graph","created_at":"2026-07-05T09:00:59Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2404.11999/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Fine-tuning pre-trained Large Language Models (LLMs) is essential to align them with human values and intentions. This process often utilizes methods like pairwise comparisons and KL divergence against a reference LLM, focusing on the evaluation of full answers generated by the models. However, the generation of these responses occurs in a token level, following a sequential, auto-regressive fashion. In this paper, we introduce Token-level Direct Preference Optimization (TDPO), a novel approach to align LLMs with human preferences by optimizing policy at the token level. Unlike previous method","authors_text":"Guoqing Liu, Haifeng Zhang, Jun Wang, Ning Yang, Weiyu Ma, Yongcheng Zeng","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-04-18T08:49:38Z","title":"Token-level Direct Preference Optimization"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.11999","kind":"arxiv","version":5},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:4d57b172fe77ef48432605250e529d5957cf6dc10c141d15cb2e7d428e804904","target":"record","created_at":"2026-07-05T09:00:59Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"dd976dfb6b2c6d8b2bbfe68ea9055987125151c152a7642ca04efae43c0996e9","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-04-18T08:49:38Z","title_canon_sha256":"c8c0399dac725367951e6eb9468b1dd4d912ba6f1d04e21f3e0447c3010abaf4"},"schema_version":"1.0","source":{"id":"2404.11999","kind":"arxiv","version":5}},"canonical_sha256":"0c4e3428902e2b4261b83a5c51b3047c0a94302027400a610ede05b88bd8065f","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"0c4e3428902e2b4261b83a5c51b3047c0a94302027400a610ede05b88bd8065f","first_computed_at":"2026-07-05T09:00:59.504850Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:00:59.504850Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"RcpSo7bSt9r0QXCsT/tnaYBJ55hbnfB/wmhQ+YxgHv3LsWtyY2S0bZxTIrM2a/CLt0MtO395u51y7Xn7Q1QiCA==","signature_status":"signed_v1","signed_at":"2026-07-05T09:00:59.505339Z","signed_message":"canonical_sha256_bytes"},"source_id":"2404.11999","source_kind":"arxiv","source_version":5}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:4d57b172fe77ef48432605250e529d5957cf6dc10c141d15cb2e7d428e804904","sha256:0bba5faa53ee4d82b8ec51d232b2c1d0f3c665ca765a9837819c31b014bdf7df"],"state_sha256":"c0b8a49a3ebc7f5aa6a7057febf81887cdff7fc584b9fd870e801e1126e2c237"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"y2ab4h1O+B8UYmfrKtKMMWNmgto1Drc6Nv4IwpddUSAGFfVMe5eMeQmx+8af2nuvF1ZKSVMehQiBCkrNQie2CQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T01:56:47.303357Z","bundle_sha256":"443cca2dffe7a6906e81d569bda8679286cd4696f574d8c3918a2274406e5cc2"}}