{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:MU77BDGIT2OL5HJMFONZARMFMM","short_pith_number":"pith:MU77BDGI","canonical_record":{"source":{"id":"2408.02666","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.CL","submitted_at":"2024-08-05T17:57:02Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"61e0375e3ac76df116e841e157a569167dec68517e09937887ac56feada5b939","abstract_canon_sha256":"0e5f3a4bd6e624992f6f59fce015e3b3d2e5c7a5164b169ffad847fb69f6a539"},"schema_version":"1.0"},"canonical_sha256":"653ff08cc89e9cbe9d2c2b9b90458563244ec61c17f580b2df133bc3ec1d8815","source":{"kind":"arxiv","id":"2408.02666","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2408.02666","created_at":"2026-07-05T08:53:25Z"},{"alias_kind":"arxiv_version","alias_value":"2408.02666v2","created_at":"2026-07-05T08:53:25Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2408.02666","created_at":"2026-07-05T08:53:25Z"},{"alias_kind":"pith_short_12","alias_value":"MU77BDGIT2OL","created_at":"2026-07-05T08:53:25Z"},{"alias_kind":"pith_short_16","alias_value":"MU77BDGIT2OL5HJM","created_at":"2026-07-05T08:53:25Z"},{"alias_kind":"pith_short_8","alias_value":"MU77BDGI","created_at":"2026-07-05T08:53:25Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:MU77BDGIT2OL5HJMFONZARMFMM","target":"record","payload":{"canonical_record":{"source":{"id":"2408.02666","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.CL","submitted_at":"2024-08-05T17:57:02Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"61e0375e3ac76df116e841e157a569167dec68517e09937887ac56feada5b939","abstract_canon_sha256":"0e5f3a4bd6e624992f6f59fce015e3b3d2e5c7a5164b169ffad847fb69f6a539"},"schema_version":"1.0"},"canonical_sha256":"653ff08cc89e9cbe9d2c2b9b90458563244ec61c17f580b2df133bc3ec1d8815","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:53:25.050782Z","signature_b64":"yHcSQAOlYQ7YfXnHg3OktNCZiZ0QWWoFDam5C5v7xOehCjhU2HGR/7RfsqObWSMkAR1BS7RdlR6rf00ll8vyDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"653ff08cc89e9cbe9d2c2b9b90458563244ec61c17f580b2df133bc3ec1d8815","last_reissued_at":"2026-07-05T08:53:25.050366Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:53:25.050366Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2408.02666","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T08:53:25Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"U94Ia6418Hd46Qrwdq7Kyae6C9S0E0yRsGIw5hg622Sdq9bg9yj17u1Lu/4C8uGkt9hRmXQOeTFfKaS88mLMCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T03:41:47.951208Z"},"content_sha256":"26e9a565d2baac562e8b6512a2a0245c061c52635893bc67c7e8c86683fd0fdb","schema_version":"1.0","event_id":"sha256:26e9a565d2baac562e8b6512a2a0245c061c52635893bc67c7e8c86683fd0fdb"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:MU77BDGIT2OL5HJMFONZARMFMM","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Self-Taught Evaluators","license":"http://creativecommons.org/publicdomain/zero/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ilia Kulikov, Jane Dwivedi-Yu, Jason Weston, Maryam Fazel-Zarandi, Olga Golovneva, Ping Yu, Richard Yuanzhe Pang, Tianlu Wang, Weizhe Yuan, Xian Li","submitted_at":"2024-08-05T17:57:02Z","abstract_excerpt":"Model-based evaluation is at the heart of successful model development -- as a reward model for training, and as a replacement for human evaluation. To train such evaluators, the standard approach is to collect a large amount of human preference judgments over model responses, which is costly and the data becomes stale as models improve. In this work, we present an approach that aims to im-prove evaluators without human annotations, using synthetic training data only. Starting from unlabeled instructions, our iterative self-improvement scheme generates contrasting model outputs and trains an L"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2408.02666","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/2408.02666/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":"2026-07-05T08:53:25Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"5FQjkFWJsJrh+P69lG4AbnHLW/nRYWzrZ0dhFMdSMRwEkS1SB8sr7QY1TBJAXD1XxEe4pa4X6ULTufzwd+nrAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T03:41:47.951730Z"},"content_sha256":"7eda086cde90348f58e416b2382adb66ebb7dbdc77718de0ad52bb2c61328d18","schema_version":"1.0","event_id":"sha256:7eda086cde90348f58e416b2382adb66ebb7dbdc77718de0ad52bb2c61328d18"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/MU77BDGIT2OL5HJMFONZARMFMM/bundle.json","state_url":"https://pith.science/pith/MU77BDGIT2OL5HJMFONZARMFMM/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/MU77BDGIT2OL5HJMFONZARMFMM/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":"2026-08-06T03:41:47Z","links":{"resolver":"https://pith.science/pith/MU77BDGIT2OL5HJMFONZARMFMM","bundle":"https://pith.science/pith/MU77BDGIT2OL5HJMFONZARMFMM/bundle.json","state":"https://pith.science/pith/MU77BDGIT2OL5HJMFONZARMFMM/state.json","well_known_bundle":"https://pith.science/.well-known/pith/MU77BDGIT2OL5HJMFONZARMFMM/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:MU77BDGIT2OL5HJMFONZARMFMM","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":"0e5f3a4bd6e624992f6f59fce015e3b3d2e5c7a5164b169ffad847fb69f6a539","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.CL","submitted_at":"2024-08-05T17:57:02Z","title_canon_sha256":"61e0375e3ac76df116e841e157a569167dec68517e09937887ac56feada5b939"},"schema_version":"1.0","source":{"id":"2408.02666","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2408.02666","created_at":"2026-07-05T08:53:25Z"},{"alias_kind":"arxiv_version","alias_value":"2408.02666v2","created_at":"2026-07-05T08:53:25Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2408.02666","created_at":"2026-07-05T08:53:25Z"},{"alias_kind":"pith_short_12","alias_value":"MU77BDGIT2OL","created_at":"2026-07-05T08:53:25Z"},{"alias_kind":"pith_short_16","alias_value":"MU77BDGIT2OL5HJM","created_at":"2026-07-05T08:53:25Z"},{"alias_kind":"pith_short_8","alias_value":"MU77BDGI","created_at":"2026-07-05T08:53:25Z"}],"graph_snapshots":[{"event_id":"sha256:7eda086cde90348f58e416b2382adb66ebb7dbdc77718de0ad52bb2c61328d18","target":"graph","created_at":"2026-07-05T08:53:25Z","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/2408.02666/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Model-based evaluation is at the heart of successful model development -- as a reward model for training, and as a replacement for human evaluation. To train such evaluators, the standard approach is to collect a large amount of human preference judgments over model responses, which is costly and the data becomes stale as models improve. In this work, we present an approach that aims to im-prove evaluators without human annotations, using synthetic training data only. Starting from unlabeled instructions, our iterative self-improvement scheme generates contrasting model outputs and trains an L","authors_text":"Ilia Kulikov, Jane Dwivedi-Yu, Jason Weston, Maryam Fazel-Zarandi, Olga Golovneva, Ping Yu, Richard Yuanzhe Pang, Tianlu Wang, Weizhe Yuan, Xian Li","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.CL","submitted_at":"2024-08-05T17:57:02Z","title":"Self-Taught Evaluators"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2408.02666","kind":"arxiv","version":2},"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":"sha256:26e9a565d2baac562e8b6512a2a0245c061c52635893bc67c7e8c86683fd0fdb","target":"record","created_at":"2026-07-05T08:53:25Z","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":"0e5f3a4bd6e624992f6f59fce015e3b3d2e5c7a5164b169ffad847fb69f6a539","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.CL","submitted_at":"2024-08-05T17:57:02Z","title_canon_sha256":"61e0375e3ac76df116e841e157a569167dec68517e09937887ac56feada5b939"},"schema_version":"1.0","source":{"id":"2408.02666","kind":"arxiv","version":2}},"canonical_sha256":"653ff08cc89e9cbe9d2c2b9b90458563244ec61c17f580b2df133bc3ec1d8815","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"653ff08cc89e9cbe9d2c2b9b90458563244ec61c17f580b2df133bc3ec1d8815","first_computed_at":"2026-07-05T08:53:25.050366Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:53:25.050366Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"yHcSQAOlYQ7YfXnHg3OktNCZiZ0QWWoFDam5C5v7xOehCjhU2HGR/7RfsqObWSMkAR1BS7RdlR6rf00ll8vyDg==","signature_status":"signed_v1","signed_at":"2026-07-05T08:53:25.050782Z","signed_message":"canonical_sha256_bytes"},"source_id":"2408.02666","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:26e9a565d2baac562e8b6512a2a0245c061c52635893bc67c7e8c86683fd0fdb","sha256:7eda086cde90348f58e416b2382adb66ebb7dbdc77718de0ad52bb2c61328d18"],"state_sha256":"67c77359c2e822cbe2a1572bc95046cf5ea71e22333243462257b800f7924739"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"GidHZGgINH72QIizy2EZpn/7V/t32oWMBc62Y5DXpCpH8j1ADWC9dz3kjQjgjQYnLdeCe7CsNbb1f1HiZjYwCQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T03:41:47.957336Z","bundle_sha256":"9e21f2cdd5deb2df6c823af415091c72435ed08a1497ac582223266ca5a94389"}}