{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:V2OKT5MH3OVVKOFUGY62YTYXLF","short_pith_number":"pith:V2OKT5MH","canonical_record":{"source":{"id":"2606.00395","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-29T22:28:08Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"123a0757d1e2396ebe14bfdfff822513bbf8b70a7a36fe25e0e6a987a45c9bd6","abstract_canon_sha256":"046b8db29b22d419488d515b855cca22a2d640b12b4316f87cdd13f3a50dc74a"},"schema_version":"1.0"},"canonical_sha256":"ae9ca9f587dbab5538b4363dac4f1759636b5ec3419d8f5f5f7ff065002b6f9f","source":{"kind":"arxiv","id":"2606.00395","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.00395","created_at":"2026-06-02T01:03:53Z"},{"alias_kind":"arxiv_version","alias_value":"2606.00395v1","created_at":"2026-06-02T01:03:53Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.00395","created_at":"2026-06-02T01:03:53Z"},{"alias_kind":"pith_short_12","alias_value":"V2OKT5MH3OVV","created_at":"2026-06-02T01:03:53Z"},{"alias_kind":"pith_short_16","alias_value":"V2OKT5MH3OVVKOFU","created_at":"2026-06-02T01:03:53Z"},{"alias_kind":"pith_short_8","alias_value":"V2OKT5MH","created_at":"2026-06-02T01:03:53Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:V2OKT5MH3OVVKOFUGY62YTYXLF","target":"record","payload":{"canonical_record":{"source":{"id":"2606.00395","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-29T22:28:08Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"123a0757d1e2396ebe14bfdfff822513bbf8b70a7a36fe25e0e6a987a45c9bd6","abstract_canon_sha256":"046b8db29b22d419488d515b855cca22a2d640b12b4316f87cdd13f3a50dc74a"},"schema_version":"1.0"},"canonical_sha256":"ae9ca9f587dbab5538b4363dac4f1759636b5ec3419d8f5f5f7ff065002b6f9f","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T01:03:53.547645Z","signature_b64":"0h40Jq6RhNRndDs6Yji1q7375G9oGCBSYm/5ze2C9jLWkwDCb6HotVCL1zEwy9AvQTRA/qmAJlQz8ymYjCdwBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ae9ca9f587dbab5538b4363dac4f1759636b5ec3419d8f5f5f7ff065002b6f9f","last_reissued_at":"2026-06-02T01:03:53.547216Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T01:03:53.547216Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2606.00395","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":"2026-06-02T01:03:53Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"JcXKFZvkoloT0s2N4VFXByGfXfbvKLrS1XRFOX9r1IBiNkmUOPQbbnTm5nQoBHeBgzQJaRxDEhytElLYsyxeDw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-06T16:00:20.499260Z"},"content_sha256":"1bd663dac8a946010429ccfc123ed81a55510605a009e28ec9bd90e405bcb017","schema_version":"1.0","event_id":"sha256:1bd663dac8a946010429ccfc123ed81a55510605a009e28ec9bd90e405bcb017"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:V2OKT5MH3OVVKOFUGY62YTYXLF","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"PR2: Predictive Routing Replay for MoE-Based LLM Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Daize Dong, Dimitris N. Metaxas, Emad Barsoum, Haolong Jia, Hongyi Wang, Huanwei Di, Jialian Wu, Jiang Liu, Jiawei Wu, Junlin Chen, Zhengzhong Liu, Zicheng Liu","submitted_at":"2026-05-29T22:28:08Z","abstract_excerpt":"Mixture of Experts (MoE) Large Language Models (LLMs) achieve strong performance at scale. However, reinforcement learning (RL) on MoE-based LLMs often suffers from training instability. A root cause is router drift, i.e., expert activations can change drastically across model updates and differ between disaggregated rollout and training phases, causing large rollout--training mismatch and unstable importance sampling weights in PPO-style RL algorithms. Routing replay mitigates this issue by freezing the replay route within each reasoning trajectory, but it ignores how the router evolves under"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.00395","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/2606.00395/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":"2026-06-02T01:03:53Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Cppibc4HFeVVU+POexAFdxzhtiODt9sQxpHuRLXrdMhHZdVWmEHtP5jo9ZeykCRBdIrO2OT847Tch+HhwsmmAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-06T16:00:20.499877Z"},"content_sha256":"39c2e22c090db8ed2960a0679630ecd6203ef5889e14613b7a6e3655a39f5008","schema_version":"1.0","event_id":"sha256:39c2e22c090db8ed2960a0679630ecd6203ef5889e14613b7a6e3655a39f5008"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/V2OKT5MH3OVVKOFUGY62YTYXLF/bundle.json","state_url":"https://pith.science/pith/V2OKT5MH3OVVKOFUGY62YTYXLF/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/V2OKT5MH3OVVKOFUGY62YTYXLF/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":"2026-06-06T16:00:20Z","links":{"resolver":"https://pith.science/pith/V2OKT5MH3OVVKOFUGY62YTYXLF","bundle":"https://pith.science/pith/V2OKT5MH3OVVKOFUGY62YTYXLF/bundle.json","state":"https://pith.science/pith/V2OKT5MH3OVVKOFUGY62YTYXLF/state.json","well_known_bundle":"https://pith.science/.well-known/pith/V2OKT5MH3OVVKOFUGY62YTYXLF/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:V2OKT5MH3OVVKOFUGY62YTYXLF","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":"046b8db29b22d419488d515b855cca22a2d640b12b4316f87cdd13f3a50dc74a","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-29T22:28:08Z","title_canon_sha256":"123a0757d1e2396ebe14bfdfff822513bbf8b70a7a36fe25e0e6a987a45c9bd6"},"schema_version":"1.0","source":{"id":"2606.00395","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.00395","created_at":"2026-06-02T01:03:53Z"},{"alias_kind":"arxiv_version","alias_value":"2606.00395v1","created_at":"2026-06-02T01:03:53Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.00395","created_at":"2026-06-02T01:03:53Z"},{"alias_kind":"pith_short_12","alias_value":"V2OKT5MH3OVV","created_at":"2026-06-02T01:03:53Z"},{"alias_kind":"pith_short_16","alias_value":"V2OKT5MH3OVVKOFU","created_at":"2026-06-02T01:03:53Z"},{"alias_kind":"pith_short_8","alias_value":"V2OKT5MH","created_at":"2026-06-02T01:03:53Z"}],"graph_snapshots":[{"event_id":"sha256:39c2e22c090db8ed2960a0679630ecd6203ef5889e14613b7a6e3655a39f5008","target":"graph","created_at":"2026-06-02T01:03:53Z","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/2606.00395/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Mixture of Experts (MoE) Large Language Models (LLMs) achieve strong performance at scale. However, reinforcement learning (RL) on MoE-based LLMs often suffers from training instability. A root cause is router drift, i.e., expert activations can change drastically across model updates and differ between disaggregated rollout and training phases, causing large rollout--training mismatch and unstable importance sampling weights in PPO-style RL algorithms. Routing replay mitigates this issue by freezing the replay route within each reasoning trajectory, but it ignores how the router evolves under","authors_text":"Daize Dong, Dimitris N. Metaxas, Emad Barsoum, Haolong Jia, Hongyi Wang, Huanwei Di, Jialian Wu, Jiang Liu, Jiawei Wu, Junlin Chen, Zhengzhong Liu, Zicheng Liu","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-29T22:28:08Z","title":"PR2: Predictive Routing Replay for MoE-Based LLM Reinforcement Learning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.00395","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":"sha256:1bd663dac8a946010429ccfc123ed81a55510605a009e28ec9bd90e405bcb017","target":"record","created_at":"2026-06-02T01:03:53Z","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":"046b8db29b22d419488d515b855cca22a2d640b12b4316f87cdd13f3a50dc74a","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-29T22:28:08Z","title_canon_sha256":"123a0757d1e2396ebe14bfdfff822513bbf8b70a7a36fe25e0e6a987a45c9bd6"},"schema_version":"1.0","source":{"id":"2606.00395","kind":"arxiv","version":1}},"canonical_sha256":"ae9ca9f587dbab5538b4363dac4f1759636b5ec3419d8f5f5f7ff065002b6f9f","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"ae9ca9f587dbab5538b4363dac4f1759636b5ec3419d8f5f5f7ff065002b6f9f","first_computed_at":"2026-06-02T01:03:53.547216Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-02T01:03:53.547216Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"0h40Jq6RhNRndDs6Yji1q7375G9oGCBSYm/5ze2C9jLWkwDCb6HotVCL1zEwy9AvQTRA/qmAJlQz8ymYjCdwBg==","signature_status":"signed_v1","signed_at":"2026-06-02T01:03:53.547645Z","signed_message":"canonical_sha256_bytes"},"source_id":"2606.00395","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:1bd663dac8a946010429ccfc123ed81a55510605a009e28ec9bd90e405bcb017","sha256:39c2e22c090db8ed2960a0679630ecd6203ef5889e14613b7a6e3655a39f5008"],"state_sha256":"6ffe159d3cf8f1ce9eaf78fb414db0c7b21dbd86576b44b249d9224f255b2b97"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"OqjHBOnCJjJDp2Ft/EvhP+d9pji+w7rlfJlJIodtyTEg3WTqB0w9LamoGonVKcgkjcE6OXgRNppbbowwM9dHAg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-06T16:00:20.502693Z","bundle_sha256":"c9fc1af07fb44ec54bf28bdabaf6b0db090cf8b50628304bb110f288ca4d56b2"}}