{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:PB6EH4S2MCUDJM7PRCWXUXC2ZB","short_pith_number":"pith:PB6EH4S2","canonical_record":{"source":{"id":"2607.15161","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-16T16:07:19Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"c9803c4df7ba3c5909eb7bf13f75ae636b2fca6b4287ad7e95388c421afb041b","abstract_canon_sha256":"56232899593f8cf3402f3acf8a13d89301b3ca89d144c80e2562edb94d1d2b9f"},"schema_version":"1.0"},"canonical_sha256":"787c43f25a60a834b3ef88ad7a5c5ac84f88c4794789fa85bb8028e199bce378","source":{"kind":"arxiv","id":"2607.15161","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2607.15161","created_at":"2026-07-17T01:22:12Z"},{"alias_kind":"arxiv_version","alias_value":"2607.15161v1","created_at":"2026-07-17T01:22:12Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.15161","created_at":"2026-07-17T01:22:12Z"},{"alias_kind":"pith_short_12","alias_value":"PB6EH4S2MCUD","created_at":"2026-07-17T01:22:12Z"},{"alias_kind":"pith_short_16","alias_value":"PB6EH4S2MCUDJM7P","created_at":"2026-07-17T01:22:12Z"},{"alias_kind":"pith_short_8","alias_value":"PB6EH4S2","created_at":"2026-07-17T01:22:12Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:PB6EH4S2MCUDJM7PRCWXUXC2ZB","target":"record","payload":{"canonical_record":{"source":{"id":"2607.15161","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-16T16:07:19Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"c9803c4df7ba3c5909eb7bf13f75ae636b2fca6b4287ad7e95388c421afb041b","abstract_canon_sha256":"56232899593f8cf3402f3acf8a13d89301b3ca89d144c80e2562edb94d1d2b9f"},"schema_version":"1.0"},"canonical_sha256":"787c43f25a60a834b3ef88ad7a5c5ac84f88c4794789fa85bb8028e199bce378","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-17T01:22:12.351630Z","signature_b64":"Uz4fzoe+FDS1s2PidAQFJUllVvcXKPuL6pD38xZc9yAlrIXCaiuLEtX4EAvrh6FF0g6WbCEpCJRGGaOarY2iAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"787c43f25a60a834b3ef88ad7a5c5ac84f88c4794789fa85bb8028e199bce378","last_reissued_at":"2026-07-17T01:22:12.350789Z","signature_status":"signed_v1","first_computed_at":"2026-07-17T01:22:12.350789Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2607.15161","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":"2026-07-17T01:22:12Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"fqLYZk5byTr9r/k5ZHFHop7vWKfEv8YJegr6aMKgmgG0ecpeV3Nf1zhpX53G+zHz6KljKz7PvnIOF60IU01vAQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-02T19:23:31.494308Z"},"content_sha256":"f0d5020485f10d19712208c0d4580c67c615c0a306c24c5a155aabb4159728fa","schema_version":"1.0","event_id":"sha256:f0d5020485f10d19712208c0d4580c67c615c0a306c24c5a155aabb4159728fa"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:PB6EH4S2MCUDJM7PRCWXUXC2ZB","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"On-Policy Delta Distillation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Byeongho Heo, Dongyoon Han, Jaehui Hwang, Sangdoo Yun","submitted_at":"2026-07-16T16:07:19Z","abstract_excerpt":"On-policy distillation is an alternative post-training method in reinforcement learning that alleviates the constraints imposed by reward models by providing token-level supervision from a teacher model. Although on-policy distillation has been studied and applied across various settings, its fundamental design remains underexplored. In this paper, we introduce a new distillation reward, termed the delta signal, instead of directly imitating the teacher's output distribution. The delta signal is defined as the difference between the teacher model and its base model prior to instruction tuning "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.15161","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/2607.15161/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":"2026-07-17T01:22:12Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"NLzD8q9Dc6EiV2JZx0CZELYUkJZKs14DpOP1luktlMPFNkYvZr+EJalztXvt5Oo6TzUc+CJujnMfh2BuoG/WBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-02T19:23:31.494998Z"},"content_sha256":"284c464ea2e962834b2dec9e0f88daa71e50118dc721a3a9b6109c206a4e2399","schema_version":"1.0","event_id":"sha256:284c464ea2e962834b2dec9e0f88daa71e50118dc721a3a9b6109c206a4e2399"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/PB6EH4S2MCUDJM7PRCWXUXC2ZB/bundle.json","state_url":"https://pith.science/pith/PB6EH4S2MCUDJM7PRCWXUXC2ZB/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/PB6EH4S2MCUDJM7PRCWXUXC2ZB/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":"2026-08-02T19:23:31Z","links":{"resolver":"https://pith.science/pith/PB6EH4S2MCUDJM7PRCWXUXC2ZB","bundle":"https://pith.science/pith/PB6EH4S2MCUDJM7PRCWXUXC2ZB/bundle.json","state":"https://pith.science/pith/PB6EH4S2MCUDJM7PRCWXUXC2ZB/state.json","well_known_bundle":"https://pith.science/.well-known/pith/PB6EH4S2MCUDJM7PRCWXUXC2ZB/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:PB6EH4S2MCUDJM7PRCWXUXC2ZB","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":"56232899593f8cf3402f3acf8a13d89301b3ca89d144c80e2562edb94d1d2b9f","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-16T16:07:19Z","title_canon_sha256":"c9803c4df7ba3c5909eb7bf13f75ae636b2fca6b4287ad7e95388c421afb041b"},"schema_version":"1.0","source":{"id":"2607.15161","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2607.15161","created_at":"2026-07-17T01:22:12Z"},{"alias_kind":"arxiv_version","alias_value":"2607.15161v1","created_at":"2026-07-17T01:22:12Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.15161","created_at":"2026-07-17T01:22:12Z"},{"alias_kind":"pith_short_12","alias_value":"PB6EH4S2MCUD","created_at":"2026-07-17T01:22:12Z"},{"alias_kind":"pith_short_16","alias_value":"PB6EH4S2MCUDJM7P","created_at":"2026-07-17T01:22:12Z"},{"alias_kind":"pith_short_8","alias_value":"PB6EH4S2","created_at":"2026-07-17T01:22:12Z"}],"graph_snapshots":[{"event_id":"sha256:284c464ea2e962834b2dec9e0f88daa71e50118dc721a3a9b6109c206a4e2399","target":"graph","created_at":"2026-07-17T01:22:12Z","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/2607.15161/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"On-policy distillation is an alternative post-training method in reinforcement learning that alleviates the constraints imposed by reward models by providing token-level supervision from a teacher model. Although on-policy distillation has been studied and applied across various settings, its fundamental design remains underexplored. In this paper, we introduce a new distillation reward, termed the delta signal, instead of directly imitating the teacher's output distribution. The delta signal is defined as the difference between the teacher model and its base model prior to instruction tuning ","authors_text":"Byeongho Heo, Dongyoon Han, Jaehui Hwang, Sangdoo Yun","cross_cats":["cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-16T16:07:19Z","title":"On-Policy Delta Distillation"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.15161","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:f0d5020485f10d19712208c0d4580c67c615c0a306c24c5a155aabb4159728fa","target":"record","created_at":"2026-07-17T01:22:12Z","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":"56232899593f8cf3402f3acf8a13d89301b3ca89d144c80e2562edb94d1d2b9f","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-16T16:07:19Z","title_canon_sha256":"c9803c4df7ba3c5909eb7bf13f75ae636b2fca6b4287ad7e95388c421afb041b"},"schema_version":"1.0","source":{"id":"2607.15161","kind":"arxiv","version":1}},"canonical_sha256":"787c43f25a60a834b3ef88ad7a5c5ac84f88c4794789fa85bb8028e199bce378","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"787c43f25a60a834b3ef88ad7a5c5ac84f88c4794789fa85bb8028e199bce378","first_computed_at":"2026-07-17T01:22:12.350789Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-17T01:22:12.350789Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"Uz4fzoe+FDS1s2PidAQFJUllVvcXKPuL6pD38xZc9yAlrIXCaiuLEtX4EAvrh6FF0g6WbCEpCJRGGaOarY2iAw==","signature_status":"signed_v1","signed_at":"2026-07-17T01:22:12.351630Z","signed_message":"canonical_sha256_bytes"},"source_id":"2607.15161","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:f0d5020485f10d19712208c0d4580c67c615c0a306c24c5a155aabb4159728fa","sha256:284c464ea2e962834b2dec9e0f88daa71e50118dc721a3a9b6109c206a4e2399"],"state_sha256":"4eb90ab5f8bae51f4ab0d0cfd8cdf0924ed169d4d5df45bc80194be05f85bc46"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"S4jw5nlO75MQEoqwE2q5L7XtX0K+mH1oNHjIYf7mzqxnu1IkTpYrsLHZQh342GIt+TGmZcQBvAulepQb6DxYCA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-02T19:23:31.501218Z","bundle_sha256":"d1652a79e8bc83181dc978657d868a66d7da6f1211b6273b9953c7eefbedde59"}}