{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:UO6GJSGJTZ5JBRAHLNP2QN73KM","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":"b904a7475888887290dcd05942d3aab071e9966588a8105e0edcce2736bcf3d3","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-10-28T14:07:41Z","title_canon_sha256":"112b2f50edd051fb8583f7c6306edbbb5b22087ce8c020fbd1d012cdc5b758eb"},"schema_version":"1.0","source":{"id":"2410.21052","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2410.21052","created_at":"2026-07-05T09:33:39Z"},{"alias_kind":"arxiv_version","alias_value":"2410.21052v3","created_at":"2026-07-05T09:33:39Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.21052","created_at":"2026-07-05T09:33:39Z"},{"alias_kind":"pith_short_12","alias_value":"UO6GJSGJTZ5J","created_at":"2026-07-05T09:33:39Z"},{"alias_kind":"pith_short_16","alias_value":"UO6GJSGJTZ5JBRAH","created_at":"2026-07-05T09:33:39Z"},{"alias_kind":"pith_short_8","alias_value":"UO6GJSGJ","created_at":"2026-07-05T09:33:39Z"}],"graph_snapshots":[{"event_id":"sha256:2ad035beebaa240322960b00cada3d6be8b8a77ca3bbb8485f0ccbcfd720b5d8","target":"graph","created_at":"2026-07-05T09:33:39Z","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/2410.21052/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"While reinforcement learning (RL) agents often perform well during training, they can struggle with distribution shift in real-world deployments. One particularly severe risk of distribution shift is goal misgeneralization, where the agent learns a proxy goal that coincides with the true goal during training but not during deployment. In this paper, we explore whether allowing an agent to ask for help from a supervisor in unfamiliar situations can mitigate this issue. We focus on agents trained with PPO in the CoinRun environment, a setting known to exhibit goal misgeneralization. We evaluate ","authors_text":"Benjamin Plaut, Mohamad H. Danesh, Nguyen X. Khanh, Tu Trinh","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-10-28T14:07:41Z","title":"Getting By Goal Misgeneralization With a Little Help From a Mentor"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.21052","kind":"arxiv","version":3},"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":"sha256:6c5baa3f6afd7435df0bca00555b24f52a828efa6e9c44ad0a9b94ef75139784","target":"record","created_at":"2026-07-05T09:33:39Z","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":"b904a7475888887290dcd05942d3aab071e9966588a8105e0edcce2736bcf3d3","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-10-28T14:07:41Z","title_canon_sha256":"112b2f50edd051fb8583f7c6306edbbb5b22087ce8c020fbd1d012cdc5b758eb"},"schema_version":"1.0","source":{"id":"2410.21052","kind":"arxiv","version":3}},"canonical_sha256":"a3bc64c8c99e7a90c4075b5fa837fb533afdd70d7198e3cf847aaafb04605626","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"a3bc64c8c99e7a90c4075b5fa837fb533afdd70d7198e3cf847aaafb04605626","first_computed_at":"2026-07-05T09:33:39.394294Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:33:39.394294Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"VwfRncsLwYZefl3ZeVLlO9gQa5CMt351FzLWy0jYhXiASyy7k7h8DAwHJko5g3gVqeWAEyW5pl5o69Fm3ufEAg==","signature_status":"signed_v1","signed_at":"2026-07-05T09:33:39.394772Z","signed_message":"canonical_sha256_bytes"},"source_id":"2410.21052","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:6c5baa3f6afd7435df0bca00555b24f52a828efa6e9c44ad0a9b94ef75139784","sha256:2ad035beebaa240322960b00cada3d6be8b8a77ca3bbb8485f0ccbcfd720b5d8"],"state_sha256":"fd7cc649dd7b58c8d53f988b7f0eeb74360daa7ee3f67d245aba457e7abe5aa4"}