{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:23J24TAOT5VNFLV5DBVY3QBCCJ","short_pith_number":"pith:23J24TAO","schema_version":"1.0","canonical_sha256":"d6d3ae4c0e9f6ad2aebd186b8dc0221279251d55586f874893ee0ee302c624e4","source":{"kind":"arxiv","id":"2203.07472","version":1},"attestation_state":"computed","paper":{"title":"Uncertainty Estimation for Language Reward Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Adam Gleave, Geoffrey Irving","submitted_at":"2022-03-14T20:13:21Z","abstract_excerpt":"Language models can learn a range of capabilities from unsupervised training on text corpora. However, to solve a particular problem (such as text summarization) it is typically necessary to fine-tune them on a task-specific dataset. It is often easier for humans to choose between options than to provide labeled data, and prior work has achieved state-of-the-art performance by training a reward model from such preference comparisons. However, collecting a large preference comparison dataset is still expensive -- and the learned reward models are unreliable out-of-distribution. We seek to addre"},"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":"2203.07472","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-03-14T20:13:21Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"d016db4a7c1ad00ad106413a5f4e6159d3892d952343578c9c1cae1fd0332eb5","abstract_canon_sha256":"445bb2579f19116d78b088a82beae5a14f08edf7223582690c1a647d1c710d8d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:05:18.841097Z","signature_b64":"ao3WhfDHuI8DTxa8aZUqCHTpBwF7N6U6pJkTXCVfwoWj6wlQcIVNS9VaK4bb5Iu5Tygw8HVAat1UHgerhwe9DQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d6d3ae4c0e9f6ad2aebd186b8dc0221279251d55586f874893ee0ee302c624e4","last_reissued_at":"2026-07-05T04:05:18.840669Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:05:18.840669Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Uncertainty Estimation for Language Reward Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Adam Gleave, Geoffrey Irving","submitted_at":"2022-03-14T20:13:21Z","abstract_excerpt":"Language models can learn a range of capabilities from unsupervised training on text corpora. However, to solve a particular problem (such as text summarization) it is typically necessary to fine-tune them on a task-specific dataset. It is often easier for humans to choose between options than to provide labeled data, and prior work has achieved state-of-the-art performance by training a reward model from such preference comparisons. However, collecting a large preference comparison dataset is still expensive -- and the learned reward models are unreliable out-of-distribution. We seek to addre"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2203.07472","kind":"arxiv","version":1},"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/2203.07472/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":"2203.07472","created_at":"2026-07-05T04:05:18.840724+00:00"},{"alias_kind":"arxiv_version","alias_value":"2203.07472v1","created_at":"2026-07-05T04:05:18.840724+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2203.07472","created_at":"2026-07-05T04:05:18.840724+00:00"},{"alias_kind":"pith_short_12","alias_value":"23J24TAOT5VN","created_at":"2026-07-05T04:05:18.840724+00:00"},{"alias_kind":"pith_short_16","alias_value":"23J24TAOT5VNFLV5","created_at":"2026-07-05T04:05:18.840724+00:00"},{"alias_kind":"pith_short_8","alias_value":"23J24TAO","created_at":"2026-07-05T04:05:18.840724+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.21602","citing_title":"Benchmarking and Improving Monitors for Out-Of-Distribution Alignment Failure in LLMs","ref_index":52,"is_internal_anchor":false},{"citing_arxiv_id":"2210.10760","citing_title":"Scaling Laws for Reward Model Overoptimization","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2605.03614","citing_title":"Uncertainty Estimation in Instance Segmentation of Affordances via Bayesian Visual Transformers","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17501","citing_title":"CoAct: Co-Active LLM Preference Learning with Human-AI Synergy","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/23J24TAOT5VNFLV5DBVY3QBCCJ","json":"https://pith.science/pith/23J24TAOT5VNFLV5DBVY3QBCCJ.json","graph_json":"https://pith.science/api/pith-number/23J24TAOT5VNFLV5DBVY3QBCCJ/graph.json","events_json":"https://pith.science/api/pith-number/23J24TAOT5VNFLV5DBVY3QBCCJ/events.json","paper":"https://pith.science/paper/23J24TAO"},"agent_actions":{"view_html":"https://pith.science/pith/23J24TAOT5VNFLV5DBVY3QBCCJ","download_json":"https://pith.science/pith/23J24TAOT5VNFLV5DBVY3QBCCJ.json","view_paper":"https://pith.science/paper/23J24TAO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2203.07472&json=true","fetch_graph":"https://pith.science/api/pith-number/23J24TAOT5VNFLV5DBVY3QBCCJ/graph.json","fetch_events":"https://pith.science/api/pith-number/23J24TAOT5VNFLV5DBVY3QBCCJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/23J24TAOT5VNFLV5DBVY3QBCCJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/23J24TAOT5VNFLV5DBVY3QBCCJ/action/storage_attestation","attest_author":"https://pith.science/pith/23J24TAOT5VNFLV5DBVY3QBCCJ/action/author_attestation","sign_citation":"https://pith.science/pith/23J24TAOT5VNFLV5DBVY3QBCCJ/action/citation_signature","submit_replication":"https://pith.science/pith/23J24TAOT5VNFLV5DBVY3QBCCJ/action/replication_record"}},"created_at":"2026-07-05T04:05:18.840724+00:00","updated_at":"2026-07-05T04:05:18.840724+00:00"}