{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:YJDU6KGMQWSUFZX5GL4R5YQ3E2","short_pith_number":"pith:YJDU6KGM","canonical_record":{"source":{"id":"2510.01857","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-10-02T09:55:26Z","cross_cats_sorted":[],"title_canon_sha256":"6a28ac5efefddb9edae47459bf6763473346f440fb319e984c41b9b6400f1388","abstract_canon_sha256":"6d314dbed37570ccc909fbb7b6ee6c4e098077bbb05c45f09759a24f32d8f5ff"},"schema_version":"1.0"},"canonical_sha256":"c2474f28cc85a542e6fd32f91ee21b26b7a68b8294cf9787d857ef88bef7d048","source":{"kind":"arxiv","id":"2510.01857","version":5},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2510.01857","created_at":"2026-05-20T00:02:56Z"},{"alias_kind":"arxiv_version","alias_value":"2510.01857v5","created_at":"2026-05-20T00:02:56Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2510.01857","created_at":"2026-05-20T00:02:56Z"},{"alias_kind":"pith_short_12","alias_value":"YJDU6KGMQWSU","created_at":"2026-05-20T00:02:56Z"},{"alias_kind":"pith_short_16","alias_value":"YJDU6KGMQWSUFZX5","created_at":"2026-05-20T00:02:56Z"},{"alias_kind":"pith_short_8","alias_value":"YJDU6KGM","created_at":"2026-05-20T00:02:56Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:YJDU6KGMQWSUFZX5GL4R5YQ3E2","target":"record","payload":{"canonical_record":{"source":{"id":"2510.01857","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-10-02T09:55:26Z","cross_cats_sorted":[],"title_canon_sha256":"6a28ac5efefddb9edae47459bf6763473346f440fb319e984c41b9b6400f1388","abstract_canon_sha256":"6d314dbed37570ccc909fbb7b6ee6c4e098077bbb05c45f09759a24f32d8f5ff"},"schema_version":"1.0"},"canonical_sha256":"c2474f28cc85a542e6fd32f91ee21b26b7a68b8294cf9787d857ef88bef7d048","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:02:56.225073Z","signature_b64":"2koTr0HwNddLULoM0hEHd11fpn4i9TmN4zbheEAfFA1u5vrY3SrSLlH09C8jgfJshb3Zcft4QGkFeVfkhzLHAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c2474f28cc85a542e6fd32f91ee21b26b7a68b8294cf9787d857ef88bef7d048","last_reissued_at":"2026-05-20T00:02:56.224220Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:02:56.224220Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2510.01857","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":"2026-05-20T00:02:56Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"V2wMIRhFVK6CEmiWpDrVx3ZHL9IlQNoSnGU6Xe/Fr2mGga2bdRH2KXG1rI8pcWfHZUKUP0U2uVUjv+dBk9zTDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-07T17:31:32.059600Z"},"content_sha256":"cf67ea7ce3ff4e235fa48117d84ae018d948d02b0cbdc69060e05ce8318c05aa","schema_version":"1.0","event_id":"sha256:cf67ea7ce3ff4e235fa48117d84ae018d948d02b0cbdc69060e05ce8318c05aa"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:YJDU6KGMQWSUFZX5GL4R5YQ3E2","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Learning Reasoning Rewards from Expert Demonstrations with Inverse Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Inverse reinforcement learning extracts reusable process rewards from expert reasoning traces that improve language model training and inference beyond imitation.","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Claudio Fanconi, Mihaela van der Schaar, Nicol\\'as Astorga","submitted_at":"2025-10-02T09:55:26Z","abstract_excerpt":"Teaching large language models (LLMs) to reason during post-training typically relies on reinforcement learning with explicit outcome- or process-based reward functions. However, in many real-world settings, obtaining or defining such reward functions is difficult, especially for complex tasks, making learning from expert demonstrations an attractive alternative. The dominant approach, supervised fine-tuning (SFT), trains models to imitate expert reasoning traces directly, but suffers from the general limitations of off-policy learning: performance can be fragile to inference-time deviations f"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Through experiments on GSM8K, MMLU-Pro and MedReason we show that the reasoning reward function learned with R-AIRL can be effectively used throughout the training and inference pipeline: (1) to provide a training signal for post-training, outperforming SFT in most of the considered settings, (2) for inference-time reranking, improving pass@1 by up to 17.4 points, and (3) for process-level evaluation, localising reasoning failures with up to 86.1% accuracy.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the adversarial inverse RL procedure can recover a generalizable process-level reward from expert demonstrations that truly reflects reasoning quality and transfers to states not explicitly present in the training traces.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"R-AIRL learns a reasoning reward function from expert demonstrations using adversarial inverse RL and shows it outperforms SFT for training while enabling reranking and failure localization on GSM8K, MMLU-Pro, and MedReason.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Inverse reinforcement learning extracts reusable process rewards from expert reasoning traces that improve language model training and inference beyond imitation.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"c9af456146c53a0d44efee218276fc18ea3f53d92fe3ec58b97efe1139437c0f"},"source":{"id":"2510.01857","kind":"arxiv","version":5},"verdict":{"id":"51d263a0-f13d-43bd-be54-1abd5bc77b61","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-18T11:08:44.647830Z","strongest_claim":"Through experiments on GSM8K, MMLU-Pro and MedReason we show that the reasoning reward function learned with R-AIRL can be effectively used throughout the training and inference pipeline: (1) to provide a training signal for post-training, outperforming SFT in most of the considered settings, (2) for inference-time reranking, improving pass@1 by up to 17.4 points, and (3) for process-level evaluation, localising reasoning failures with up to 86.1% accuracy.","one_line_summary":"R-AIRL learns a reasoning reward function from expert demonstrations using adversarial inverse RL and shows it outperforms SFT for training while enabling reranking and failure localization on GSM8K, MMLU-Pro, and MedReason.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the adversarial inverse RL procedure can recover a generalizable process-level reward from expert demonstrations that truly reflects reasoning quality and transfers to states not explicitly present in the training traces.","pith_extraction_headline":"Inverse reinforcement learning extracts reusable process rewards from expert reasoning traces that improve language model training and inference beyond imitation."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2510.01857/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":1,"snapshot_sha256":"be01e8ccc0eb607e8627d867ccb4d8e809ece5b944f26d198b08080121140881"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"51d263a0-f13d-43bd-be54-1abd5bc77b61"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T00:02:56Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"oSjGuEsf5A2MmuWzgwoPqO2KK13PHG98CntZW0UxhsfIW/qf6EuOg6aSQU2lwYSvjYkrGR+w9HhdzaiHGjqoBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-07T17:31:32.060489Z"},"content_sha256":"848196b061248c68f629fa46d01a3d1e8f5298cbca3dd82c6a711691487ae8c5","schema_version":"1.0","event_id":"sha256:848196b061248c68f629fa46d01a3d1e8f5298cbca3dd82c6a711691487ae8c5"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/YJDU6KGMQWSUFZX5GL4R5YQ3E2/bundle.json","state_url":"https://pith.science/pith/YJDU6KGMQWSUFZX5GL4R5YQ3E2/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/YJDU6KGMQWSUFZX5GL4R5YQ3E2/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":"2026-06-07T17:31:32Z","links":{"resolver":"https://pith.science/pith/YJDU6KGMQWSUFZX5GL4R5YQ3E2","bundle":"https://pith.science/pith/YJDU6KGMQWSUFZX5GL4R5YQ3E2/bundle.json","state":"https://pith.science/pith/YJDU6KGMQWSUFZX5GL4R5YQ3E2/state.json","well_known_bundle":"https://pith.science/.well-known/pith/YJDU6KGMQWSUFZX5GL4R5YQ3E2/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:YJDU6KGMQWSUFZX5GL4R5YQ3E2","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":"6d314dbed37570ccc909fbb7b6ee6c4e098077bbb05c45f09759a24f32d8f5ff","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-10-02T09:55:26Z","title_canon_sha256":"6a28ac5efefddb9edae47459bf6763473346f440fb319e984c41b9b6400f1388"},"schema_version":"1.0","source":{"id":"2510.01857","kind":"arxiv","version":5}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2510.01857","created_at":"2026-05-20T00:02:56Z"},{"alias_kind":"arxiv_version","alias_value":"2510.01857v5","created_at":"2026-05-20T00:02:56Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2510.01857","created_at":"2026-05-20T00:02:56Z"},{"alias_kind":"pith_short_12","alias_value":"YJDU6KGMQWSU","created_at":"2026-05-20T00:02:56Z"},{"alias_kind":"pith_short_16","alias_value":"YJDU6KGMQWSUFZX5","created_at":"2026-05-20T00:02:56Z"},{"alias_kind":"pith_short_8","alias_value":"YJDU6KGM","created_at":"2026-05-20T00:02:56Z"}],"graph_snapshots":[{"event_id":"sha256:848196b061248c68f629fa46d01a3d1e8f5298cbca3dd82c6a711691487ae8c5","target":"graph","created_at":"2026-05-20T00:02:56Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Through experiments on GSM8K, MMLU-Pro and MedReason we show that the reasoning reward function learned with R-AIRL can be effectively used throughout the training and inference pipeline: (1) to provide a training signal for post-training, outperforming SFT in most of the considered settings, (2) for inference-time reranking, improving pass@1 by up to 17.4 points, and (3) for process-level evaluation, localising reasoning failures with up to 86.1% accuracy."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the adversarial inverse RL procedure can recover a generalizable process-level reward from expert demonstrations that truly reflects reasoning quality and transfers to states not explicitly present in the training traces."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"R-AIRL learns a reasoning reward function from expert demonstrations using adversarial inverse RL and shows it outperforms SFT for training while enabling reranking and failure localization on GSM8K, MMLU-Pro, and MedReason."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Inverse reinforcement learning extracts reusable process rewards from expert reasoning traces that improve language model training and inference beyond imitation."}],"snapshot_sha256":"c9af456146c53a0d44efee218276fc18ea3f53d92fe3ec58b97efe1139437c0f"},"formal_canon":{"evidence_count":1,"snapshot_sha256":"be01e8ccc0eb607e8627d867ccb4d8e809ece5b944f26d198b08080121140881"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2510.01857/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Teaching large language models (LLMs) to reason during post-training typically relies on reinforcement learning with explicit outcome- or process-based reward functions. However, in many real-world settings, obtaining or defining such reward functions is difficult, especially for complex tasks, making learning from expert demonstrations an attractive alternative. The dominant approach, supervised fine-tuning (SFT), trains models to imitate expert reasoning traces directly, but suffers from the general limitations of off-policy learning: performance can be fragile to inference-time deviations f","authors_text":"Claudio Fanconi, Mihaela van der Schaar, Nicol\\'as Astorga","cross_cats":[],"headline":"Inverse reinforcement learning extracts reusable process rewards from expert reasoning traces that improve language model training and inference beyond imitation.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-10-02T09:55:26Z","title":"Learning Reasoning Rewards from Expert Demonstrations with Inverse Reinforcement Learning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2510.01857","kind":"arxiv","version":5},"verdict":{"created_at":"2026-05-18T11:08:44.647830Z","id":"51d263a0-f13d-43bd-be54-1abd5bc77b61","model_set":{"reader":"grok-4.3"},"one_line_summary":"R-AIRL learns a reasoning reward function from expert demonstrations using adversarial inverse RL and shows it outperforms SFT for training while enabling reranking and failure localization on GSM8K, MMLU-Pro, and MedReason.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Inverse reinforcement learning extracts reusable process rewards from expert reasoning traces that improve language model training and inference beyond imitation.","strongest_claim":"Through experiments on GSM8K, MMLU-Pro and MedReason we show that the reasoning reward function learned with R-AIRL can be effectively used throughout the training and inference pipeline: (1) to provide a training signal for post-training, outperforming SFT in most of the considered settings, (2) for inference-time reranking, improving pass@1 by up to 17.4 points, and (3) for process-level evaluation, localising reasoning failures with up to 86.1% accuracy.","weakest_assumption":"That the adversarial inverse RL procedure can recover a generalizable process-level reward from expert demonstrations that truly reflects reasoning quality and transfers to states not explicitly present in the training traces."}},"verdict_id":"51d263a0-f13d-43bd-be54-1abd5bc77b61"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:cf67ea7ce3ff4e235fa48117d84ae018d948d02b0cbdc69060e05ce8318c05aa","target":"record","created_at":"2026-05-20T00:02:56Z","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":"6d314dbed37570ccc909fbb7b6ee6c4e098077bbb05c45f09759a24f32d8f5ff","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-10-02T09:55:26Z","title_canon_sha256":"6a28ac5efefddb9edae47459bf6763473346f440fb319e984c41b9b6400f1388"},"schema_version":"1.0","source":{"id":"2510.01857","kind":"arxiv","version":5}},"canonical_sha256":"c2474f28cc85a542e6fd32f91ee21b26b7a68b8294cf9787d857ef88bef7d048","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"c2474f28cc85a542e6fd32f91ee21b26b7a68b8294cf9787d857ef88bef7d048","first_computed_at":"2026-05-20T00:02:56.224220Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-20T00:02:56.224220Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"2koTr0HwNddLULoM0hEHd11fpn4i9TmN4zbheEAfFA1u5vrY3SrSLlH09C8jgfJshb3Zcft4QGkFeVfkhzLHAQ==","signature_status":"signed_v1","signed_at":"2026-05-20T00:02:56.225073Z","signed_message":"canonical_sha256_bytes"},"source_id":"2510.01857","source_kind":"arxiv","source_version":5}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:cf67ea7ce3ff4e235fa48117d84ae018d948d02b0cbdc69060e05ce8318c05aa","sha256:848196b061248c68f629fa46d01a3d1e8f5298cbca3dd82c6a711691487ae8c5"],"state_sha256":"3f0dffd466ec2fca89c7c6fbd33d49b4088829b468a382a67a540030cc306305"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"MxSOtBtTY2fDTUT4MbngzPyOh52DzE3XLjNnHCn8aO24EunMSkW/+ZoznC7HWCkd6RATv3X3ZWdggL2Mr9eIAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-07T17:31:32.065241Z","bundle_sha256":"ed6cc825752c4a94b6fb8fa5bb7043f26b2beffa2a91ae8dda6f1610775cb794"}}