{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:SR4SNW3AAQVRBVENIKIIB5WAYR","short_pith_number":"pith:SR4SNW3A","canonical_record":{"source":{"id":"2605.12655","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-05-12T19:01:16Z","cross_cats_sorted":["cs.MA"],"title_canon_sha256":"43132de58f7a27c4c4df01db7fb108f2740977bcc8221c99ca27420f797786e6","abstract_canon_sha256":"13f9e0b6287753427bb2b64460e41c6fe500857010a03b3d0ec76c17189af15d"},"schema_version":"1.0"},"canonical_sha256":"947926db60042b10d48d429080f6c0c4639ef9fc6b35e048df11e7fe2a900836","source":{"kind":"arxiv","id":"2605.12655","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.12655","created_at":"2026-05-18T03:09:50Z"},{"alias_kind":"arxiv_version","alias_value":"2605.12655v1","created_at":"2026-05-18T03:09:50Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.12655","created_at":"2026-05-18T03:09:50Z"},{"alias_kind":"pith_short_12","alias_value":"SR4SNW3AAQVR","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"SR4SNW3AAQVRBVEN","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"SR4SNW3A","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:SR4SNW3AAQVRBVENIKIIB5WAYR","target":"record","payload":{"canonical_record":{"source":{"id":"2605.12655","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-05-12T19:01:16Z","cross_cats_sorted":["cs.MA"],"title_canon_sha256":"43132de58f7a27c4c4df01db7fb108f2740977bcc8221c99ca27420f797786e6","abstract_canon_sha256":"13f9e0b6287753427bb2b64460e41c6fe500857010a03b3d0ec76c17189af15d"},"schema_version":"1.0"},"canonical_sha256":"947926db60042b10d48d429080f6c0c4639ef9fc6b35e048df11e7fe2a900836","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T03:09:50.661982Z","signature_b64":"ZmVB8uY7OwQykk5LC/55UDBsKmKWhWqO4JwoQfNmLVpvkBygky+O/eqhFQzO3/Qejx1V4IdgBUczPw5DinbbBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"947926db60042b10d48d429080f6c0c4639ef9fc6b35e048df11e7fe2a900836","last_reissued_at":"2026-05-18T03:09:50.661167Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T03:09:50.661167Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.12655","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":"2026-05-18T03:09:50Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"jQPOYz85tGvR3hiXlHJpvjURB0yMGZtfNCzXQ3EL86n26JwzZQQHQh1Ry6DKnmq07WvHMhwIr6soS05wVDlgCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-31T15:26:50.709890Z"},"content_sha256":"7a9e2a1d9cbac91fbc406aacd9540c2233932f8cebd39c7f105ca71cc2c4a72d","schema_version":"1.0","event_id":"sha256:7a9e2a1d9cbac91fbc406aacd9540c2233932f8cebd39c7f105ca71cc2c4a72d"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:SR4SNW3AAQVRBVENIKIIB5WAYR","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Macro-Action Based Multi-Agent Instruction Following through Value Cancellation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Correcting the Bellman backup target at each instruction boundary decouples value estimates across contexts, allowing a single policy to follow interrupting instructions while preserving base-task performance in multi-agent settings.","cross_cats":["cs.MA"],"primary_cat":"cs.AI","authors_text":"Enrico Marchesini Xiang Zhi Tan, Ethan Rathbun, Wo Wei Lin","submitted_at":"2026-05-12T19:01:16Z","abstract_excerpt":"Multi-agent reinforcement learning (MARL) in real-world use cases may need to adapt to external natural language instructions that interrupt ongoing behavior and conflict with long-horizon objectives. However, conditioning rewards on instructions introduces a fundamental failure mode as Bellman updates couple value estimates across instruction contexts, leading to inconsistent values when instructions interrupt macro-actions. We propose Macro-Action Value Correction for Instruction Compliance (MAVIC), which corrects Bellman backups at instruction boundaries by correcting the incoming instructi"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"MAVIC achieves high instruction compliance while preserving base task performance in increasingly complex cooperative multi-agent environments.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That correcting the bootstrapping target at instruction boundaries is sufficient to fully decouple value estimates across contexts without introducing new inconsistencies under stochastic switching or macro-action interruptions.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"MAVIC corrects Bellman backups at instruction boundaries by adjusting the incoming objective and restoring continuation value, enabling consistent estimation under stochastic instruction switching in a unified policy.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Correcting the Bellman backup target at each instruction boundary decouples value estimates across contexts, allowing a single policy to follow interrupting instructions while preserving base-task performance in multi-agent settings.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"db942b2271ccb422000eb5a68eb94c0fce7bf00258e2963a03f50d84c618cd04"},"source":{"id":"2605.12655","kind":"arxiv","version":1},"verdict":{"id":"7d79834b-368f-4ab7-815a-1cdca8494a7b","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T20:32:37.115796Z","strongest_claim":"MAVIC achieves high instruction compliance while preserving base task performance in increasingly complex cooperative multi-agent environments.","one_line_summary":"MAVIC corrects Bellman backups at instruction boundaries by adjusting the incoming objective and restoring continuation value, enabling consistent estimation under stochastic instruction switching in a unified policy.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That correcting the bootstrapping target at instruction boundaries is sufficient to fully decouple value estimates across contexts without introducing new inconsistencies under stochastic switching or macro-action interruptions.","pith_extraction_headline":"Correcting the Bellman backup target at each instruction boundary decouples value estimates across contexts, allowing a single policy to follow interrupting instructions while preserving base-task performance in multi-agent settings."},"references":{"count":4,"sample":[{"doi":"","year":2021,"title":"bring me the tomato,","work_id":"c529cb9f-0686-4e65-9c98-3a674afd2fcb","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1999,"title":"When c=c ′, the reward is unchanged","work_id":"b7590f98-9c18-416a-bcc3-e45947033d44","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Don’t use left cutting board","work_id":"adad8480-b299-45e0-8d41-eafaaff87697","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Go to small boxes","work_id":"ced46404-1005-42e0-8a0d-ba22a2e91f4c","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":4,"snapshot_sha256":"dc6afdeced95dd3b205a114113da17d69f20013332fe56e6db4f3b2b1d7a6f4d","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":"7d79834b-368f-4ab7-815a-1cdca8494a7b"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T03:09:50Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Jy++W2HUNcsbQ2OAFWx1lk3vX5Vgv5rn+V+rZY+vsslij6u5KmSjyKxj+LV24eb02FaAMVOsI16POT+qCPpKCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-31T15:26:50.710941Z"},"content_sha256":"64f561269c6487139150067eb84685ea0b98ceeb502130dfba852ed05d1b78fc","schema_version":"1.0","event_id":"sha256:64f561269c6487139150067eb84685ea0b98ceeb502130dfba852ed05d1b78fc"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/SR4SNW3AAQVRBVENIKIIB5WAYR/bundle.json","state_url":"https://pith.science/pith/SR4SNW3AAQVRBVENIKIIB5WAYR/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/SR4SNW3AAQVRBVENIKIIB5WAYR/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":"2026-05-31T15:26:50Z","links":{"resolver":"https://pith.science/pith/SR4SNW3AAQVRBVENIKIIB5WAYR","bundle":"https://pith.science/pith/SR4SNW3AAQVRBVENIKIIB5WAYR/bundle.json","state":"https://pith.science/pith/SR4SNW3AAQVRBVENIKIIB5WAYR/state.json","well_known_bundle":"https://pith.science/.well-known/pith/SR4SNW3AAQVRBVENIKIIB5WAYR/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:SR4SNW3AAQVRBVENIKIIB5WAYR","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":"13f9e0b6287753427bb2b64460e41c6fe500857010a03b3d0ec76c17189af15d","cross_cats_sorted":["cs.MA"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-05-12T19:01:16Z","title_canon_sha256":"43132de58f7a27c4c4df01db7fb108f2740977bcc8221c99ca27420f797786e6"},"schema_version":"1.0","source":{"id":"2605.12655","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.12655","created_at":"2026-05-18T03:09:50Z"},{"alias_kind":"arxiv_version","alias_value":"2605.12655v1","created_at":"2026-05-18T03:09:50Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.12655","created_at":"2026-05-18T03:09:50Z"},{"alias_kind":"pith_short_12","alias_value":"SR4SNW3AAQVR","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"SR4SNW3AAQVRBVEN","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"SR4SNW3A","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:64f561269c6487139150067eb84685ea0b98ceeb502130dfba852ed05d1b78fc","target":"graph","created_at":"2026-05-18T03:09:50Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"MAVIC achieves high instruction compliance while preserving base task performance in increasingly complex cooperative multi-agent environments."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That correcting the bootstrapping target at instruction boundaries is sufficient to fully decouple value estimates across contexts without introducing new inconsistencies under stochastic switching or macro-action interruptions."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"MAVIC corrects Bellman backups at instruction boundaries by adjusting the incoming objective and restoring continuation value, enabling consistent estimation under stochastic instruction switching in a unified policy."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Correcting the Bellman backup target at each instruction boundary decouples value estimates across contexts, allowing a single policy to follow interrupting instructions while preserving base-task performance in multi-agent settings."}],"snapshot_sha256":"db942b2271ccb422000eb5a68eb94c0fce7bf00258e2963a03f50d84c618cd04"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"Multi-agent reinforcement learning (MARL) in real-world use cases may need to adapt to external natural language instructions that interrupt ongoing behavior and conflict with long-horizon objectives. However, conditioning rewards on instructions introduces a fundamental failure mode as Bellman updates couple value estimates across instruction contexts, leading to inconsistent values when instructions interrupt macro-actions. We propose Macro-Action Value Correction for Instruction Compliance (MAVIC), which corrects Bellman backups at instruction boundaries by correcting the incoming instructi","authors_text":"Enrico Marchesini Xiang Zhi Tan, Ethan Rathbun, Wo Wei Lin","cross_cats":["cs.MA"],"headline":"Correcting the Bellman backup target at each instruction boundary decouples value estimates across contexts, allowing a single policy to follow interrupting instructions while preserving base-task performance in multi-agent settings.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-05-12T19:01:16Z","title":"Macro-Action Based Multi-Agent Instruction Following through Value Cancellation"},"references":{"count":4,"internal_anchors":0,"resolved_work":4,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"bring me the tomato,","work_id":"c529cb9f-0686-4e65-9c98-3a674afd2fcb","year":2021},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"When c=c ′, the reward is unchanged","work_id":"b7590f98-9c18-416a-bcc3-e45947033d44","year":1999},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Don’t use left cutting board","work_id":"adad8480-b299-45e0-8d41-eafaaff87697","year":2022},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Go to small boxes","work_id":"ced46404-1005-42e0-8a0d-ba22a2e91f4c","year":2022}],"snapshot_sha256":"dc6afdeced95dd3b205a114113da17d69f20013332fe56e6db4f3b2b1d7a6f4d"},"source":{"id":"2605.12655","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-14T20:32:37.115796Z","id":"7d79834b-368f-4ab7-815a-1cdca8494a7b","model_set":{"reader":"grok-4.3"},"one_line_summary":"MAVIC corrects Bellman backups at instruction boundaries by adjusting the incoming objective and restoring continuation value, enabling consistent estimation under stochastic instruction switching in a unified policy.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Correcting the Bellman backup target at each instruction boundary decouples value estimates across contexts, allowing a single policy to follow interrupting instructions while preserving base-task performance in multi-agent settings.","strongest_claim":"MAVIC achieves high instruction compliance while preserving base task performance in increasingly complex cooperative multi-agent environments.","weakest_assumption":"That correcting the bootstrapping target at instruction boundaries is sufficient to fully decouple value estimates across contexts without introducing new inconsistencies under stochastic switching or macro-action interruptions."}},"verdict_id":"7d79834b-368f-4ab7-815a-1cdca8494a7b"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:7a9e2a1d9cbac91fbc406aacd9540c2233932f8cebd39c7f105ca71cc2c4a72d","target":"record","created_at":"2026-05-18T03:09:50Z","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":"13f9e0b6287753427bb2b64460e41c6fe500857010a03b3d0ec76c17189af15d","cross_cats_sorted":["cs.MA"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-05-12T19:01:16Z","title_canon_sha256":"43132de58f7a27c4c4df01db7fb108f2740977bcc8221c99ca27420f797786e6"},"schema_version":"1.0","source":{"id":"2605.12655","kind":"arxiv","version":1}},"canonical_sha256":"947926db60042b10d48d429080f6c0c4639ef9fc6b35e048df11e7fe2a900836","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"947926db60042b10d48d429080f6c0c4639ef9fc6b35e048df11e7fe2a900836","first_computed_at":"2026-05-18T03:09:50.661167Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T03:09:50.661167Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"ZmVB8uY7OwQykk5LC/55UDBsKmKWhWqO4JwoQfNmLVpvkBygky+O/eqhFQzO3/Qejx1V4IdgBUczPw5DinbbBQ==","signature_status":"signed_v1","signed_at":"2026-05-18T03:09:50.661982Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.12655","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:7a9e2a1d9cbac91fbc406aacd9540c2233932f8cebd39c7f105ca71cc2c4a72d","sha256:64f561269c6487139150067eb84685ea0b98ceeb502130dfba852ed05d1b78fc"],"state_sha256":"7ebc8bc02f6619d3d9659a62d2dfaf36a887f3c84eb1fcad66703aca81a717f2"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"B0x2u+7rpuK6U7rOZysXvl3zO3jPBoR19RvL0Q2ehIIoVCj3XIyRtJdHrKOL7/7WdHCbBmpIJy3exjzfZFiJCw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-31T15:26:50.715594Z","bundle_sha256":"2c3f523858b4538997f43a3fd80c4ddc1bdca99815553dd53d2b3e98fcf98758"}}