{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:XQW5AIMXSI4WCRD2KM4FR3TNML","short_pith_number":"pith:XQW5AIMX","canonical_record":{"source":{"id":"2606.09821","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-08T17:58:23Z","cross_cats_sorted":[],"title_canon_sha256":"dd093fc1d6cd7ab56c3ea1905269a61b09d5767173dd3bb52397cf4c36116d72","abstract_canon_sha256":"82d9ffc9ac850e3b1a4722068eb2441c2768b63a9b0eb5ece39df29de369c1ba"},"schema_version":"1.0"},"canonical_sha256":"bc2dd02197923961447a533858ee6d62f2a510e3d1af153e3b2094288f4eb4c3","source":{"kind":"arxiv","id":"2606.09821","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.09821","created_at":"2026-06-09T02:09:12Z"},{"alias_kind":"arxiv_version","alias_value":"2606.09821v1","created_at":"2026-06-09T02:09:12Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.09821","created_at":"2026-06-09T02:09:12Z"},{"alias_kind":"pith_short_12","alias_value":"XQW5AIMXSI4W","created_at":"2026-06-09T02:09:12Z"},{"alias_kind":"pith_short_16","alias_value":"XQW5AIMXSI4WCRD2","created_at":"2026-06-09T02:09:12Z"},{"alias_kind":"pith_short_8","alias_value":"XQW5AIMX","created_at":"2026-06-09T02:09:12Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:XQW5AIMXSI4WCRD2KM4FR3TNML","target":"record","payload":{"canonical_record":{"source":{"id":"2606.09821","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-08T17:58:23Z","cross_cats_sorted":[],"title_canon_sha256":"dd093fc1d6cd7ab56c3ea1905269a61b09d5767173dd3bb52397cf4c36116d72","abstract_canon_sha256":"82d9ffc9ac850e3b1a4722068eb2441c2768b63a9b0eb5ece39df29de369c1ba"},"schema_version":"1.0"},"canonical_sha256":"bc2dd02197923961447a533858ee6d62f2a510e3d1af153e3b2094288f4eb4c3","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T02:09:12.426206Z","signature_b64":"is771FhLFx2XXczGhkR5/njA6+5VyakNBNL4F8gD5DBk826uDOs5+G4fL0IccWB9yYAzbXKY2bc7ZxUnsyHcCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bc2dd02197923961447a533858ee6d62f2a510e3d1af153e3b2094288f4eb4c3","last_reissued_at":"2026-06-09T02:09:12.425341Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T02:09:12.425341Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2606.09821","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":"2026-06-09T02:09:12Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"WZfRz7TGlfewr/ry/4zmWsXI8JhQZfBqtPm9JdXdelnHrt1/h11H9xICXR0824a4ElIubCKmjYE5nEwtR2PrBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-27T20:27:08.902545Z"},"content_sha256":"7934158f22c29363237e12461821807d6d30b8ad0885c0ac7b63306514c21bde","schema_version":"1.0","event_id":"sha256:7934158f22c29363237e12461821807d6d30b8ad0885c0ac7b63306514c21bde"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:XQW5AIMXSI4WCRD2KM4FR3TNML","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Rethinking the Divergence Regularization in LLM RL","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Jiarui Yao, Liefeng Bo, Penghui Qi, Tianyu Pang, Wee Sun Lee, Xiangxin Zhou","submitted_at":"2026-06-08T17:58:23Z","abstract_excerpt":"Reinforcement learning (RL) has become a key component of post-training large language models (LLMs). In practice, LLM RL is often off-policy because of training-inference mismatch and policy staleness, making trust-region control essential for stable optimization. Mainstream methods such as PPO and GRPO approximate this control with a ratio-clipping mechanism, but the importance ratio can be a poor proxy for distributional shift in long-tailed vocabularies. Recent work such as DPPO addresses this mismatch by replacing ratio-based clipping with a divergence-based mask, yielding a trust region "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.09821","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/2606.09821/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":"2026-06-09T02:09:12Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"6pCzCz8TTR1y73Ip/w9LNqmOwJS00O+r1lgLPWcD4Q6dQjfbGAJ02iRxT1qpoAIEK9NS+OycDA+kd4tEjfJSDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-27T20:27:08.902904Z"},"content_sha256":"741322cb0dc5298870d2bdc4fc0ebaf9c55e2fcf5e40e69323a4bd0edd4fb891","schema_version":"1.0","event_id":"sha256:741322cb0dc5298870d2bdc4fc0ebaf9c55e2fcf5e40e69323a4bd0edd4fb891"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/XQW5AIMXSI4WCRD2KM4FR3TNML/bundle.json","state_url":"https://pith.science/pith/XQW5AIMXSI4WCRD2KM4FR3TNML/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/XQW5AIMXSI4WCRD2KM4FR3TNML/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":"2026-06-27T20:27:08Z","links":{"resolver":"https://pith.science/pith/XQW5AIMXSI4WCRD2KM4FR3TNML","bundle":"https://pith.science/pith/XQW5AIMXSI4WCRD2KM4FR3TNML/bundle.json","state":"https://pith.science/pith/XQW5AIMXSI4WCRD2KM4FR3TNML/state.json","well_known_bundle":"https://pith.science/.well-known/pith/XQW5AIMXSI4WCRD2KM4FR3TNML/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:XQW5AIMXSI4WCRD2KM4FR3TNML","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":"82d9ffc9ac850e3b1a4722068eb2441c2768b63a9b0eb5ece39df29de369c1ba","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-08T17:58:23Z","title_canon_sha256":"dd093fc1d6cd7ab56c3ea1905269a61b09d5767173dd3bb52397cf4c36116d72"},"schema_version":"1.0","source":{"id":"2606.09821","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.09821","created_at":"2026-06-09T02:09:12Z"},{"alias_kind":"arxiv_version","alias_value":"2606.09821v1","created_at":"2026-06-09T02:09:12Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.09821","created_at":"2026-06-09T02:09:12Z"},{"alias_kind":"pith_short_12","alias_value":"XQW5AIMXSI4W","created_at":"2026-06-09T02:09:12Z"},{"alias_kind":"pith_short_16","alias_value":"XQW5AIMXSI4WCRD2","created_at":"2026-06-09T02:09:12Z"},{"alias_kind":"pith_short_8","alias_value":"XQW5AIMX","created_at":"2026-06-09T02:09:12Z"}],"graph_snapshots":[{"event_id":"sha256:741322cb0dc5298870d2bdc4fc0ebaf9c55e2fcf5e40e69323a4bd0edd4fb891","target":"graph","created_at":"2026-06-09T02:09:12Z","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/2606.09821/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Reinforcement learning (RL) has become a key component of post-training large language models (LLMs). In practice, LLM RL is often off-policy because of training-inference mismatch and policy staleness, making trust-region control essential for stable optimization. Mainstream methods such as PPO and GRPO approximate this control with a ratio-clipping mechanism, but the importance ratio can be a poor proxy for distributional shift in long-tailed vocabularies. Recent work such as DPPO addresses this mismatch by replacing ratio-based clipping with a divergence-based mask, yielding a trust region ","authors_text":"Jiarui Yao, Liefeng Bo, Penghui Qi, Tianyu Pang, Wee Sun Lee, Xiangxin Zhou","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-08T17:58:23Z","title":"Rethinking the Divergence Regularization in LLM RL"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.09821","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":"sha256:7934158f22c29363237e12461821807d6d30b8ad0885c0ac7b63306514c21bde","target":"record","created_at":"2026-06-09T02:09:12Z","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":"82d9ffc9ac850e3b1a4722068eb2441c2768b63a9b0eb5ece39df29de369c1ba","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-08T17:58:23Z","title_canon_sha256":"dd093fc1d6cd7ab56c3ea1905269a61b09d5767173dd3bb52397cf4c36116d72"},"schema_version":"1.0","source":{"id":"2606.09821","kind":"arxiv","version":1}},"canonical_sha256":"bc2dd02197923961447a533858ee6d62f2a510e3d1af153e3b2094288f4eb4c3","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"bc2dd02197923961447a533858ee6d62f2a510e3d1af153e3b2094288f4eb4c3","first_computed_at":"2026-06-09T02:09:12.425341Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-09T02:09:12.425341Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"is771FhLFx2XXczGhkR5/njA6+5VyakNBNL4F8gD5DBk826uDOs5+G4fL0IccWB9yYAzbXKY2bc7ZxUnsyHcCA==","signature_status":"signed_v1","signed_at":"2026-06-09T02:09:12.426206Z","signed_message":"canonical_sha256_bytes"},"source_id":"2606.09821","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:7934158f22c29363237e12461821807d6d30b8ad0885c0ac7b63306514c21bde","sha256:741322cb0dc5298870d2bdc4fc0ebaf9c55e2fcf5e40e69323a4bd0edd4fb891"],"state_sha256":"322894d8a5767b235f7bc8fd11605e5984e394e3e7d10a18ce4dd1b92810732e"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"aLMOkxD2Qkzfhhgtd7U6YeoMAXvPo7Spo8BAoBsvAda8VUTQXpeEMiS8hU9UakCNyGeD35CXV7eOGFjZDvBXDg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-27T20:27:08.904969Z","bundle_sha256":"103c2e9edd0db5aa4d213965c663bc12e06cd560715f2a538cb1df4ef00b8ce4"}}