{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2019:LDOSFEOGVFXXJI5HS57ZRTOYXC","short_pith_number":"pith:LDOSFEOG","canonical_record":{"source":{"id":"1907.01475","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-07-02T16:12:34Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"59349a745f4160954b0c1e9573b65e6fdf684c109574c151d11ef935425d1576","abstract_canon_sha256":"44bdf09742f1cd8ca8cc89e0ca8b63007461d6eaa4a6994bcc45e3e50b66180f"},"schema_version":"1.0"},"canonical_sha256":"58dd2291c6a96f74a3a7977f98cdd8b8b6bbf3a5f7a0b8a815d6668fca83005c","source":{"kind":"arxiv","id":"1907.01475","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1907.01475","created_at":"2026-05-17T23:41:40Z"},{"alias_kind":"arxiv_version","alias_value":"1907.01475v1","created_at":"2026-05-17T23:41:40Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1907.01475","created_at":"2026-05-17T23:41:40Z"},{"alias_kind":"pith_short_12","alias_value":"LDOSFEOGVFXX","created_at":"2026-05-18T12:33:21Z"},{"alias_kind":"pith_short_16","alias_value":"LDOSFEOGVFXXJI5H","created_at":"2026-05-18T12:33:21Z"},{"alias_kind":"pith_short_8","alias_value":"LDOSFEOG","created_at":"2026-05-18T12:33:21Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2019:LDOSFEOGVFXXJI5HS57ZRTOYXC","target":"record","payload":{"canonical_record":{"source":{"id":"1907.01475","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-07-02T16:12:34Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"59349a745f4160954b0c1e9573b65e6fdf684c109574c151d11ef935425d1576","abstract_canon_sha256":"44bdf09742f1cd8ca8cc89e0ca8b63007461d6eaa4a6994bcc45e3e50b66180f"},"schema_version":"1.0"},"canonical_sha256":"58dd2291c6a96f74a3a7977f98cdd8b8b6bbf3a5f7a0b8a815d6668fca83005c","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:41:40.297808Z","signature_b64":"hsqAIPCfXPD59tb4ARCQf80r2TyAQ/Z28JBnOpIJhNwEG30MSQzNqSlg3G4nHpOFPBJCQH++6oVocJQ/jQA+BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"58dd2291c6a96f74a3a7977f98cdd8b8b6bbf3a5f7a0b8a815d6668fca83005c","last_reissued_at":"2026-05-17T23:41:40.297328Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:41:40.297328Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"1907.01475","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":"2026-05-17T23:41:40Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"H5dKbHVvoSslSUGY8+LFvEx+Zt8RB8ha3SwcUydw0lPvy+ECCxiVK+Xi9UaXuhIiARMOhgwp/H2MYkxitJtQDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-10T22:27:50.314975Z"},"content_sha256":"ebea1de41be240209aa09cdebd2696d5ba4e560892cce55d960ac108ea8a7e04","schema_version":"1.0","event_id":"sha256:ebea1de41be240209aa09cdebd2696d5ba4e560892cce55d960ac108ea8a7e04"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2019:LDOSFEOGVFXXJI5HS57ZRTOYXC","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Generalizing from a few environments in safety-critical reinforcement learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Angelos Filos, Owain Evans, Yarin Gal, Zachary Kenton","submitted_at":"2019-07-02T16:12:34Z","abstract_excerpt":"Before deploying autonomous agents in the real world, we need to be confident they will perform safely in novel situations. Ideally, we would expose agents to a very wide range of situations during training, allowing them to learn about every possible danger, but this is often impractical. This paper investigates safety and generalization from a limited number of training environments in deep reinforcement learning (RL). We find RL algorithms can fail dangerously on unseen test environments even when performing perfectly on training environments. Firstly, in a gridworld setting, we show that c"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1907.01475","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":""},"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":"2026-05-17T23:41:40Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"2JvIHW7Uq4T+tdpAsE9oRv1/Ymt/E0mo1XeW/buk+fA2wuMDP1ZEI7Dt3I8naGzWm2fqZiklUCHzsEtMg8/ZDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-10T22:27:50.315907Z"},"content_sha256":"7810080d259e244ff6d6ac0a210756e202bd74467fda8da9c8e5d9df71816446","schema_version":"1.0","event_id":"sha256:7810080d259e244ff6d6ac0a210756e202bd74467fda8da9c8e5d9df71816446"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/LDOSFEOGVFXXJI5HS57ZRTOYXC/bundle.json","state_url":"https://pith.science/pith/LDOSFEOGVFXXJI5HS57ZRTOYXC/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/LDOSFEOGVFXXJI5HS57ZRTOYXC/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":"2026-06-10T22:27:50Z","links":{"resolver":"https://pith.science/pith/LDOSFEOGVFXXJI5HS57ZRTOYXC","bundle":"https://pith.science/pith/LDOSFEOGVFXXJI5HS57ZRTOYXC/bundle.json","state":"https://pith.science/pith/LDOSFEOGVFXXJI5HS57ZRTOYXC/state.json","well_known_bundle":"https://pith.science/.well-known/pith/LDOSFEOGVFXXJI5HS57ZRTOYXC/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2019:LDOSFEOGVFXXJI5HS57ZRTOYXC","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":"44bdf09742f1cd8ca8cc89e0ca8b63007461d6eaa4a6994bcc45e3e50b66180f","cross_cats_sorted":["cs.AI","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-07-02T16:12:34Z","title_canon_sha256":"59349a745f4160954b0c1e9573b65e6fdf684c109574c151d11ef935425d1576"},"schema_version":"1.0","source":{"id":"1907.01475","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1907.01475","created_at":"2026-05-17T23:41:40Z"},{"alias_kind":"arxiv_version","alias_value":"1907.01475v1","created_at":"2026-05-17T23:41:40Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1907.01475","created_at":"2026-05-17T23:41:40Z"},{"alias_kind":"pith_short_12","alias_value":"LDOSFEOGVFXX","created_at":"2026-05-18T12:33:21Z"},{"alias_kind":"pith_short_16","alias_value":"LDOSFEOGVFXXJI5H","created_at":"2026-05-18T12:33:21Z"},{"alias_kind":"pith_short_8","alias_value":"LDOSFEOG","created_at":"2026-05-18T12:33:21Z"}],"graph_snapshots":[{"event_id":"sha256:7810080d259e244ff6d6ac0a210756e202bd74467fda8da9c8e5d9df71816446","target":"graph","created_at":"2026-05-17T23:41:40Z","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"},"paper":{"abstract_excerpt":"Before deploying autonomous agents in the real world, we need to be confident they will perform safely in novel situations. Ideally, we would expose agents to a very wide range of situations during training, allowing them to learn about every possible danger, but this is often impractical. This paper investigates safety and generalization from a limited number of training environments in deep reinforcement learning (RL). We find RL algorithms can fail dangerously on unseen test environments even when performing perfectly on training environments. Firstly, in a gridworld setting, we show that c","authors_text":"Angelos Filos, Owain Evans, Yarin Gal, Zachary Kenton","cross_cats":["cs.AI","stat.ML"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-07-02T16:12:34Z","title":"Generalizing from a few environments in safety-critical reinforcement learning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1907.01475","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:ebea1de41be240209aa09cdebd2696d5ba4e560892cce55d960ac108ea8a7e04","target":"record","created_at":"2026-05-17T23:41:40Z","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":"44bdf09742f1cd8ca8cc89e0ca8b63007461d6eaa4a6994bcc45e3e50b66180f","cross_cats_sorted":["cs.AI","stat.ML"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-07-02T16:12:34Z","title_canon_sha256":"59349a745f4160954b0c1e9573b65e6fdf684c109574c151d11ef935425d1576"},"schema_version":"1.0","source":{"id":"1907.01475","kind":"arxiv","version":1}},"canonical_sha256":"58dd2291c6a96f74a3a7977f98cdd8b8b6bbf3a5f7a0b8a815d6668fca83005c","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"58dd2291c6a96f74a3a7977f98cdd8b8b6bbf3a5f7a0b8a815d6668fca83005c","first_computed_at":"2026-05-17T23:41:40.297328Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:41:40.297328Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"hsqAIPCfXPD59tb4ARCQf80r2TyAQ/Z28JBnOpIJhNwEG30MSQzNqSlg3G4nHpOFPBJCQH++6oVocJQ/jQA+BQ==","signature_status":"signed_v1","signed_at":"2026-05-17T23:41:40.297808Z","signed_message":"canonical_sha256_bytes"},"source_id":"1907.01475","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:ebea1de41be240209aa09cdebd2696d5ba4e560892cce55d960ac108ea8a7e04","sha256:7810080d259e244ff6d6ac0a210756e202bd74467fda8da9c8e5d9df71816446"],"state_sha256":"838ce05a1c7ace886aff9bcddd3f3ce20bdbb59f362828297cfe3b2834b039ba"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"+Ucr2K4pGazbgqQAAf0XngNdBXE8JpEq2s0HOjGZkK42Ri2hmSond3cdZ38fTuV0Zcxogr+CercpMjzOeOjEAQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-10T22:27:50.320214Z","bundle_sha256":"0058001d53aa873fdb169ae06026245bafe37142ced2660626a643d7e26c8dde"}}