{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:HNTXAGV5VUEERYS3UFPQ2G4P5L","short_pith_number":"pith:HNTXAGV5","canonical_record":{"source":{"id":"2606.08446","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-07T04:24:45Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"2b302eb64bb662dc8245267ec91c6138a9bcb45bcc853def10ca9cd108a0b5f3","abstract_canon_sha256":"bf5ef923f92ada9dbed10caecddba30504cef770c5096c1a3200533ba5a05c2f"},"schema_version":"1.0"},"canonical_sha256":"3b67701abdad0848e25ba15f0d1b8feaf79dd2bdb8517261e07f18558279319a","source":{"kind":"arxiv","id":"2606.08446","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.08446","created_at":"2026-06-09T01:05:36Z"},{"alias_kind":"arxiv_version","alias_value":"2606.08446v1","created_at":"2026-06-09T01:05:36Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.08446","created_at":"2026-06-09T01:05:36Z"},{"alias_kind":"pith_short_12","alias_value":"HNTXAGV5VUEE","created_at":"2026-06-09T01:05:36Z"},{"alias_kind":"pith_short_16","alias_value":"HNTXAGV5VUEERYS3","created_at":"2026-06-09T01:05:36Z"},{"alias_kind":"pith_short_8","alias_value":"HNTXAGV5","created_at":"2026-06-09T01:05:36Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:HNTXAGV5VUEERYS3UFPQ2G4P5L","target":"record","payload":{"canonical_record":{"source":{"id":"2606.08446","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-07T04:24:45Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"2b302eb64bb662dc8245267ec91c6138a9bcb45bcc853def10ca9cd108a0b5f3","abstract_canon_sha256":"bf5ef923f92ada9dbed10caecddba30504cef770c5096c1a3200533ba5a05c2f"},"schema_version":"1.0"},"canonical_sha256":"3b67701abdad0848e25ba15f0d1b8feaf79dd2bdb8517261e07f18558279319a","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T01:05:36.867528Z","signature_b64":"Fk+3I0BkVT58hNCD06FWeD+0H0Ku8T5bK9atUtWXL//T7TlcKKxFl1wUhPRbE4qmfs6v1J17PN7SjY5mERYhAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3b67701abdad0848e25ba15f0d1b8feaf79dd2bdb8517261e07f18558279319a","last_reissued_at":"2026-06-09T01:05:36.867141Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T01:05:36.867141Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2606.08446","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":"2026-06-09T01:05:36Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"l7bWmv6r9y2BpsVha5Qkz1wjlweFbB68SF6kjYWdG/Zz00ClesMKGQDQlXheyw2o20+MC1VDIF/Ioz4hwaqkDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-12T03:40:00.659950Z"},"content_sha256":"c688b7f11ffa7278359566de2eec04122ad630b457cee4c29b7952129eddd8e8","schema_version":"1.0","event_id":"sha256:c688b7f11ffa7278359566de2eec04122ad630b457cee4c29b7952129eddd8e8"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:HNTXAGV5VUEERYS3UFPQ2G4P5L","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Sparrow: Sparse Rollout for Stable and Efficient Long-context RL of Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Aram Galstyan, Beidi Chen, Haizhong Zheng, Ranajoy Sadhukhan, Sai Muralidhar Jayanthi, Saket Dingliwal, Souvik Kundu, Yang Zhou, Zhaofeng Sun, Zhuoming Chen","submitted_at":"2026-06-07T04:24:45Z","abstract_excerpt":"Despite being powerful, reinforcement learning with verifiable rewards (RLVR) induces extremely long COT, making it computationally expensive. Since RLVR per-step cost is dominated by long-context rollout generation, sparse attention offers a promising way to accelerate dense rollout. However, sparse rollouts require a delicate stability-efficiency tradeoff: overly aggressive sparsity causes collapse, while overly lenient sparsity gives insufficient speedup. In this work, we study this tradeoff through sparse-to-dense actor-policy mismatch. We first observe that sparse rollout collapse is not "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.08446","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/2606.08446/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":"2026-06-09T01:05:36Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"a7IXAiebRkbgIpIMbc4Cyt8L+J8z0WiWwBC2sU6Kne2M2N1zhkBp4osMEbZRRJR+yV9Q3ZHelxuV60Ycb7bCBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-12T03:40:00.660863Z"},"content_sha256":"6d6c5d2520fc9a8c671a47478447a95144dc89a6737c12483c52c50d985e9635","schema_version":"1.0","event_id":"sha256:6d6c5d2520fc9a8c671a47478447a95144dc89a6737c12483c52c50d985e9635"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/HNTXAGV5VUEERYS3UFPQ2G4P5L/bundle.json","state_url":"https://pith.science/pith/HNTXAGV5VUEERYS3UFPQ2G4P5L/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/HNTXAGV5VUEERYS3UFPQ2G4P5L/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":"2026-06-12T03:40:00Z","links":{"resolver":"https://pith.science/pith/HNTXAGV5VUEERYS3UFPQ2G4P5L","bundle":"https://pith.science/pith/HNTXAGV5VUEERYS3UFPQ2G4P5L/bundle.json","state":"https://pith.science/pith/HNTXAGV5VUEERYS3UFPQ2G4P5L/state.json","well_known_bundle":"https://pith.science/.well-known/pith/HNTXAGV5VUEERYS3UFPQ2G4P5L/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:HNTXAGV5VUEERYS3UFPQ2G4P5L","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":"bf5ef923f92ada9dbed10caecddba30504cef770c5096c1a3200533ba5a05c2f","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-07T04:24:45Z","title_canon_sha256":"2b302eb64bb662dc8245267ec91c6138a9bcb45bcc853def10ca9cd108a0b5f3"},"schema_version":"1.0","source":{"id":"2606.08446","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.08446","created_at":"2026-06-09T01:05:36Z"},{"alias_kind":"arxiv_version","alias_value":"2606.08446v1","created_at":"2026-06-09T01:05:36Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.08446","created_at":"2026-06-09T01:05:36Z"},{"alias_kind":"pith_short_12","alias_value":"HNTXAGV5VUEE","created_at":"2026-06-09T01:05:36Z"},{"alias_kind":"pith_short_16","alias_value":"HNTXAGV5VUEERYS3","created_at":"2026-06-09T01:05:36Z"},{"alias_kind":"pith_short_8","alias_value":"HNTXAGV5","created_at":"2026-06-09T01:05:36Z"}],"graph_snapshots":[{"event_id":"sha256:6d6c5d2520fc9a8c671a47478447a95144dc89a6737c12483c52c50d985e9635","target":"graph","created_at":"2026-06-09T01:05:36Z","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/2606.08446/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Despite being powerful, reinforcement learning with verifiable rewards (RLVR) induces extremely long COT, making it computationally expensive. Since RLVR per-step cost is dominated by long-context rollout generation, sparse attention offers a promising way to accelerate dense rollout. However, sparse rollouts require a delicate stability-efficiency tradeoff: overly aggressive sparsity causes collapse, while overly lenient sparsity gives insufficient speedup. In this work, we study this tradeoff through sparse-to-dense actor-policy mismatch. We first observe that sparse rollout collapse is not ","authors_text":"Aram Galstyan, Beidi Chen, Haizhong Zheng, Ranajoy Sadhukhan, Sai Muralidhar Jayanthi, Saket Dingliwal, Souvik Kundu, Yang Zhou, Zhaofeng Sun, Zhuoming Chen","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-07T04:24:45Z","title":"Sparrow: Sparse Rollout for Stable and Efficient Long-context RL of Large Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.08446","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:c688b7f11ffa7278359566de2eec04122ad630b457cee4c29b7952129eddd8e8","target":"record","created_at":"2026-06-09T01:05:36Z","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":"bf5ef923f92ada9dbed10caecddba30504cef770c5096c1a3200533ba5a05c2f","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-07T04:24:45Z","title_canon_sha256":"2b302eb64bb662dc8245267ec91c6138a9bcb45bcc853def10ca9cd108a0b5f3"},"schema_version":"1.0","source":{"id":"2606.08446","kind":"arxiv","version":1}},"canonical_sha256":"3b67701abdad0848e25ba15f0d1b8feaf79dd2bdb8517261e07f18558279319a","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"3b67701abdad0848e25ba15f0d1b8feaf79dd2bdb8517261e07f18558279319a","first_computed_at":"2026-06-09T01:05:36.867141Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-09T01:05:36.867141Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"Fk+3I0BkVT58hNCD06FWeD+0H0Ku8T5bK9atUtWXL//T7TlcKKxFl1wUhPRbE4qmfs6v1J17PN7SjY5mERYhAA==","signature_status":"signed_v1","signed_at":"2026-06-09T01:05:36.867528Z","signed_message":"canonical_sha256_bytes"},"source_id":"2606.08446","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c688b7f11ffa7278359566de2eec04122ad630b457cee4c29b7952129eddd8e8","sha256:6d6c5d2520fc9a8c671a47478447a95144dc89a6737c12483c52c50d985e9635"],"state_sha256":"d14f22c6d19f20427e1da4ff897c13350c104fcd23ad5678e7e0e168a8862a75"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"8JDAemKvQ3CVdlgSseSzlH9Rk7T//rMsb6en0fzGi6uOOPR9N2R6R6URom3QNkdoVlgEw/N34wZT+atZ+zg1Aw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-12T03:40:00.665695Z","bundle_sha256":"d7fc23f6b2fad619f424e48f694f11dcb445de6d833bd67bec5e3d5aeb7fb948"}}