{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:GNDKEPQJQ5W4VYZGT27HMVBFJB","short_pith_number":"pith:GNDKEPQJ","canonical_record":{"source":{"id":"2605.30232","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-28T17:03:18Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"bba937eb41467eef6cf6818b997a495e1b7b2b4b678e833bdb121e1130a03eaf","abstract_canon_sha256":"b7a4c0ffc5cce1e128f1a9cae084a622b3bad2455631bc67dcf7e7abe7e172e8"},"schema_version":"1.0"},"canonical_sha256":"3346a23e09876dcae3269ebe765425487568f8723f548882eb2eab0503a93001","source":{"kind":"arxiv","id":"2605.30232","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.30232","created_at":"2026-05-29T02:06:13Z"},{"alias_kind":"arxiv_version","alias_value":"2605.30232v1","created_at":"2026-05-29T02:06:13Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.30232","created_at":"2026-05-29T02:06:13Z"},{"alias_kind":"pith_short_12","alias_value":"GNDKEPQJQ5W4","created_at":"2026-05-29T02:06:13Z"},{"alias_kind":"pith_short_16","alias_value":"GNDKEPQJQ5W4VYZG","created_at":"2026-05-29T02:06:13Z"},{"alias_kind":"pith_short_8","alias_value":"GNDKEPQJ","created_at":"2026-05-29T02:06:13Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:GNDKEPQJQ5W4VYZGT27HMVBFJB","target":"record","payload":{"canonical_record":{"source":{"id":"2605.30232","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-28T17:03:18Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"bba937eb41467eef6cf6818b997a495e1b7b2b4b678e833bdb121e1130a03eaf","abstract_canon_sha256":"b7a4c0ffc5cce1e128f1a9cae084a622b3bad2455631bc67dcf7e7abe7e172e8"},"schema_version":"1.0"},"canonical_sha256":"3346a23e09876dcae3269ebe765425487568f8723f548882eb2eab0503a93001","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-29T02:06:13.502581Z","signature_b64":"7ke7SCTFnG92/ku/vDDIwBL5aRKGlTvpS7j4gAvVFyf6/Lg6X5oALBb+HWxgd1qF2ccUKvVHP7yhlXTve4zXDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3346a23e09876dcae3269ebe765425487568f8723f548882eb2eab0503a93001","last_reissued_at":"2026-05-29T02:06:13.502211Z","signature_status":"signed_v1","first_computed_at":"2026-05-29T02:06:13.502211Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.30232","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":"2026-05-29T02:06:13Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Sw6XiJRRoHolos6u0dCWS1klGQGa+rbgVop/ktezWvtDteTDXTIqU5sjDdl1zh6U0uEI6vYJpL2Pfd850/zuCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-29T16:04:54.345107Z"},"content_sha256":"fa7e235c5281b2b8720c21ce9e8118a66255e074d3662c390d79c9da0cd9fcf2","schema_version":"1.0","event_id":"sha256:fa7e235c5281b2b8720c21ce9e8118a66255e074d3662c390d79c9da0cd9fcf2"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:GNDKEPQJQ5W4VYZGT27HMVBFJB","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"How's it going? Reinforcement learning in language models recruits a functional welfare axis","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Andy Q Han, David J. Chalmers, Pavel Izmailov","submitted_at":"2026-05-28T17:03:18Z","abstract_excerpt":"How does reinforcement learning shape a language model's internal representations? We present evidence that RL recruits a pre-existing representation of functional welfare: an estimate of how well or badly the system is doing, relative to its goals. We train several language models in a novel, semantically neutral maze environment. We then extract concept vectors for rewarded and punished trajectories, and evaluate those vectors in settings unrelated to the maze environment. The punishment vector behaves like a representation of negative welfare: it promotes failure and impossibility tokens, i"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.30232","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/2605.30232/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":"2026-05-29T02:06:13Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"JyDHrout+YqnAb3wx78tiLtkprPt60ofFZDWpEfXFUlmjiHObeKZzeRegJEo4H62NE1rPt+2+09HGDmrmRUGBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-29T16:04:54.345499Z"},"content_sha256":"8928af5313075bd1ba4710bfc850a5e38fe1c3983f696603ff5b19c9bbb8736d","schema_version":"1.0","event_id":"sha256:8928af5313075bd1ba4710bfc850a5e38fe1c3983f696603ff5b19c9bbb8736d"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/GNDKEPQJQ5W4VYZGT27HMVBFJB/bundle.json","state_url":"https://pith.science/pith/GNDKEPQJQ5W4VYZGT27HMVBFJB/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/GNDKEPQJQ5W4VYZGT27HMVBFJB/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":"2026-06-29T16:04:54Z","links":{"resolver":"https://pith.science/pith/GNDKEPQJQ5W4VYZGT27HMVBFJB","bundle":"https://pith.science/pith/GNDKEPQJQ5W4VYZGT27HMVBFJB/bundle.json","state":"https://pith.science/pith/GNDKEPQJQ5W4VYZGT27HMVBFJB/state.json","well_known_bundle":"https://pith.science/.well-known/pith/GNDKEPQJQ5W4VYZGT27HMVBFJB/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:GNDKEPQJQ5W4VYZGT27HMVBFJB","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":"b7a4c0ffc5cce1e128f1a9cae084a622b3bad2455631bc67dcf7e7abe7e172e8","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-28T17:03:18Z","title_canon_sha256":"bba937eb41467eef6cf6818b997a495e1b7b2b4b678e833bdb121e1130a03eaf"},"schema_version":"1.0","source":{"id":"2605.30232","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.30232","created_at":"2026-05-29T02:06:13Z"},{"alias_kind":"arxiv_version","alias_value":"2605.30232v1","created_at":"2026-05-29T02:06:13Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.30232","created_at":"2026-05-29T02:06:13Z"},{"alias_kind":"pith_short_12","alias_value":"GNDKEPQJQ5W4","created_at":"2026-05-29T02:06:13Z"},{"alias_kind":"pith_short_16","alias_value":"GNDKEPQJQ5W4VYZG","created_at":"2026-05-29T02:06:13Z"},{"alias_kind":"pith_short_8","alias_value":"GNDKEPQJ","created_at":"2026-05-29T02:06:13Z"}],"graph_snapshots":[{"event_id":"sha256:8928af5313075bd1ba4710bfc850a5e38fe1c3983f696603ff5b19c9bbb8736d","target":"graph","created_at":"2026-05-29T02:06:13Z","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/2605.30232/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"How does reinforcement learning shape a language model's internal representations? We present evidence that RL recruits a pre-existing representation of functional welfare: an estimate of how well or badly the system is doing, relative to its goals. We train several language models in a novel, semantically neutral maze environment. We then extract concept vectors for rewarded and punished trajectories, and evaluate those vectors in settings unrelated to the maze environment. The punishment vector behaves like a representation of negative welfare: it promotes failure and impossibility tokens, i","authors_text":"Andy Q Han, David J. Chalmers, Pavel Izmailov","cross_cats":["cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-28T17:03:18Z","title":"How's it going? Reinforcement learning in language models recruits a functional welfare axis"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.30232","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:fa7e235c5281b2b8720c21ce9e8118a66255e074d3662c390d79c9da0cd9fcf2","target":"record","created_at":"2026-05-29T02:06:13Z","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":"b7a4c0ffc5cce1e128f1a9cae084a622b3bad2455631bc67dcf7e7abe7e172e8","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-28T17:03:18Z","title_canon_sha256":"bba937eb41467eef6cf6818b997a495e1b7b2b4b678e833bdb121e1130a03eaf"},"schema_version":"1.0","source":{"id":"2605.30232","kind":"arxiv","version":1}},"canonical_sha256":"3346a23e09876dcae3269ebe765425487568f8723f548882eb2eab0503a93001","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"3346a23e09876dcae3269ebe765425487568f8723f548882eb2eab0503a93001","first_computed_at":"2026-05-29T02:06:13.502211Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-29T02:06:13.502211Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"7ke7SCTFnG92/ku/vDDIwBL5aRKGlTvpS7j4gAvVFyf6/Lg6X5oALBb+HWxgd1qF2ccUKvVHP7yhlXTve4zXDw==","signature_status":"signed_v1","signed_at":"2026-05-29T02:06:13.502581Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.30232","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:fa7e235c5281b2b8720c21ce9e8118a66255e074d3662c390d79c9da0cd9fcf2","sha256:8928af5313075bd1ba4710bfc850a5e38fe1c3983f696603ff5b19c9bbb8736d"],"state_sha256":"fc92a97721809e57bd74004da667cdfab5a45590e70e7a0ffadca2286b7dbd86"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"u5plKxYNncAPVPdyPUyiEqwV01Pdz7ts5qHw6v+dQiW/DWOqfMQ1FV8vKymEKAXeCN+PUiU6ZQQCl4h0SgY/CQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-29T16:04:54.347386Z","bundle_sha256":"8ad444824c5340ef06e8a3ad30d1f54f245e315b1623500b7a3c624bd4c13d92"}}