{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:GCXMS5FMJGOEADOLGIEF7L5WP3","short_pith_number":"pith:GCXMS5FM","canonical_record":{"source":{"id":"2605.27140","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-26T15:07:03Z","cross_cats_sorted":[],"title_canon_sha256":"c1f2c0acf38b62c797832abf981d4e64fdc32f9d57ce4811ba7c6aa76eeee8fe","abstract_canon_sha256":"1fc30d83f73a9df1241075095b9eef29297bbbe73bd0948a98e307f8a5863300"},"schema_version":"1.0"},"canonical_sha256":"30aec974ac499c400dcb32085fafb67ed7de1155967977ee8ae489cb7fc9584a","source":{"kind":"arxiv","id":"2605.27140","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.27140","created_at":"2026-05-27T02:05:43Z"},{"alias_kind":"arxiv_version","alias_value":"2605.27140v1","created_at":"2026-05-27T02:05:43Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.27140","created_at":"2026-05-27T02:05:43Z"},{"alias_kind":"pith_short_12","alias_value":"GCXMS5FMJGOE","created_at":"2026-05-27T02:05:43Z"},{"alias_kind":"pith_short_16","alias_value":"GCXMS5FMJGOEADOL","created_at":"2026-05-27T02:05:43Z"},{"alias_kind":"pith_short_8","alias_value":"GCXMS5FM","created_at":"2026-05-27T02:05:43Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:GCXMS5FMJGOEADOLGIEF7L5WP3","target":"record","payload":{"canonical_record":{"source":{"id":"2605.27140","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-26T15:07:03Z","cross_cats_sorted":[],"title_canon_sha256":"c1f2c0acf38b62c797832abf981d4e64fdc32f9d57ce4811ba7c6aa76eeee8fe","abstract_canon_sha256":"1fc30d83f73a9df1241075095b9eef29297bbbe73bd0948a98e307f8a5863300"},"schema_version":"1.0"},"canonical_sha256":"30aec974ac499c400dcb32085fafb67ed7de1155967977ee8ae489cb7fc9584a","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-27T02:05:43.876359Z","signature_b64":"dnLlq3E0dojYyrxJxsnQzTUvf18V76etTvew/qypvlNZ+HgM5Sq1/0iUfaVvGPBEjBCum5OakTvH/t0Qee5oBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"30aec974ac499c400dcb32085fafb67ed7de1155967977ee8ae489cb7fc9584a","last_reissued_at":"2026-05-27T02:05:43.875677Z","signature_status":"signed_v1","first_computed_at":"2026-05-27T02:05:43.875677Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.27140","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-27T02:05:43Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"gTtK5P0Kd3bHthWVJHoNB0namL7bCjbugUbpJ20b4jR8sCG6Himh7Du/XkS8DSb81V1RHX4Bm0dL2jbrdvzRCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-31T16:05:07.012759Z"},"content_sha256":"a49b167a519b940b9abbdeaea4f62eac6686acb37af8a8d0bbc13e21e9c89004","schema_version":"1.0","event_id":"sha256:a49b167a519b940b9abbdeaea4f62eac6686acb37af8a8d0bbc13e21e9c89004"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:GCXMS5FMJGOEADOLGIEF7L5WP3","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"StepOPSD: Step-Aware Online Preference Distillation for Agent Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Chenglin Wu, Xu Lin, Yanfei Zhang","submitted_at":"2026-05-26T15:07:03Z","abstract_excerpt":"Reinforcement learning for multi-turn agents suffers from a credit-assignment mismatch: rewards are sparse and trajectory-level, while success often hinges on a few local decisions. Existing online policy distillation (OPD) provides denser token-level supervision, but typically treats heterogeneous agent trajectories as monolithic strings rather than causal interaction units. We present StepOPSD, a post-rollout preference self-distillation framework that takes the agent step as the unit of credit redistribution. StepOPSD decomposes trajectories into action-centered step segments, rescoring the"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.27140","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/2605.27140/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":"2026-05-27T02:05:43Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"SALaRHjin8fnN47ZBiv46cpNFw4jFubvJe/4wcfDZ7vF/tcYv1gG94eZXSD5w2oxL0WZ46+1j4R84YiPPyBWDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-31T16:05:07.013394Z"},"content_sha256":"a0090fe2352c888cc7823dfd41461a644a43c103d7e50554706b4f759b41fda9","schema_version":"1.0","event_id":"sha256:a0090fe2352c888cc7823dfd41461a644a43c103d7e50554706b4f759b41fda9"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/GCXMS5FMJGOEADOLGIEF7L5WP3/bundle.json","state_url":"https://pith.science/pith/GCXMS5FMJGOEADOLGIEF7L5WP3/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/GCXMS5FMJGOEADOLGIEF7L5WP3/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":"2026-05-31T16:05:07Z","links":{"resolver":"https://pith.science/pith/GCXMS5FMJGOEADOLGIEF7L5WP3","bundle":"https://pith.science/pith/GCXMS5FMJGOEADOLGIEF7L5WP3/bundle.json","state":"https://pith.science/pith/GCXMS5FMJGOEADOLGIEF7L5WP3/state.json","well_known_bundle":"https://pith.science/.well-known/pith/GCXMS5FMJGOEADOLGIEF7L5WP3/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:GCXMS5FMJGOEADOLGIEF7L5WP3","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":"1fc30d83f73a9df1241075095b9eef29297bbbe73bd0948a98e307f8a5863300","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-26T15:07:03Z","title_canon_sha256":"c1f2c0acf38b62c797832abf981d4e64fdc32f9d57ce4811ba7c6aa76eeee8fe"},"schema_version":"1.0","source":{"id":"2605.27140","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.27140","created_at":"2026-05-27T02:05:43Z"},{"alias_kind":"arxiv_version","alias_value":"2605.27140v1","created_at":"2026-05-27T02:05:43Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.27140","created_at":"2026-05-27T02:05:43Z"},{"alias_kind":"pith_short_12","alias_value":"GCXMS5FMJGOE","created_at":"2026-05-27T02:05:43Z"},{"alias_kind":"pith_short_16","alias_value":"GCXMS5FMJGOEADOL","created_at":"2026-05-27T02:05:43Z"},{"alias_kind":"pith_short_8","alias_value":"GCXMS5FM","created_at":"2026-05-27T02:05:43Z"}],"graph_snapshots":[{"event_id":"sha256:a0090fe2352c888cc7823dfd41461a644a43c103d7e50554706b4f759b41fda9","target":"graph","created_at":"2026-05-27T02:05:43Z","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/2605.27140/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Reinforcement learning for multi-turn agents suffers from a credit-assignment mismatch: rewards are sparse and trajectory-level, while success often hinges on a few local decisions. Existing online policy distillation (OPD) provides denser token-level supervision, but typically treats heterogeneous agent trajectories as monolithic strings rather than causal interaction units. We present StepOPSD, a post-rollout preference self-distillation framework that takes the agent step as the unit of credit redistribution. StepOPSD decomposes trajectories into action-centered step segments, rescoring the","authors_text":"Chenglin Wu, Xu Lin, Yanfei Zhang","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-26T15:07:03Z","title":"StepOPSD: Step-Aware Online Preference Distillation for Agent Reinforcement Learning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.27140","kind":"arxiv","version":1},"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:a49b167a519b940b9abbdeaea4f62eac6686acb37af8a8d0bbc13e21e9c89004","target":"record","created_at":"2026-05-27T02:05:43Z","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":"1fc30d83f73a9df1241075095b9eef29297bbbe73bd0948a98e307f8a5863300","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-26T15:07:03Z","title_canon_sha256":"c1f2c0acf38b62c797832abf981d4e64fdc32f9d57ce4811ba7c6aa76eeee8fe"},"schema_version":"1.0","source":{"id":"2605.27140","kind":"arxiv","version":1}},"canonical_sha256":"30aec974ac499c400dcb32085fafb67ed7de1155967977ee8ae489cb7fc9584a","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"30aec974ac499c400dcb32085fafb67ed7de1155967977ee8ae489cb7fc9584a","first_computed_at":"2026-05-27T02:05:43.875677Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-27T02:05:43.875677Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"dnLlq3E0dojYyrxJxsnQzTUvf18V76etTvew/qypvlNZ+HgM5Sq1/0iUfaVvGPBEjBCum5OakTvH/t0Qee5oBQ==","signature_status":"signed_v1","signed_at":"2026-05-27T02:05:43.876359Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.27140","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:a49b167a519b940b9abbdeaea4f62eac6686acb37af8a8d0bbc13e21e9c89004","sha256:a0090fe2352c888cc7823dfd41461a644a43c103d7e50554706b4f759b41fda9"],"state_sha256":"5917b514ffee9106e2d333a0f9f377ec71f5878aa0d68ea350efb91748ba13ba"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"CIVObViareqADI+sb1Wdnhr3MedTlhezOsqAc9y0dmwIK2SSKyyhcAXv2cjbA2YH0IPVtnyv8+CVy9dGyQyfAw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-31T16:05:07.017381Z","bundle_sha256":"f00cd8ce27fe750a3cc6ee2a59e51c706bd58e3b8f3efeb3539878838ffd51cc"}}