{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:E47PO7DKKC6L4TXC722UOYQWYD","short_pith_number":"pith:E47PO7DK","canonical_record":{"source":{"id":"2606.13209","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-11T11:19:03Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"00793d84fef45565cf4a29d94d7fc2f91518d84b5e545caf5ddacdc29806f9d4","abstract_canon_sha256":"38a4aa43ee0c482cd8359e4cb2f19cc5baf3af1556bd16839fa03b30f8790762"},"schema_version":"1.0"},"canonical_sha256":"273ef77c6a50bcbe4ee2feb5476216c0d1c259843e4f140c3c02f580a299085d","source":{"kind":"arxiv","id":"2606.13209","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.13209","created_at":"2026-06-12T01:09:46Z"},{"alias_kind":"arxiv_version","alias_value":"2606.13209v1","created_at":"2026-06-12T01:09:46Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.13209","created_at":"2026-06-12T01:09:46Z"},{"alias_kind":"pith_short_12","alias_value":"E47PO7DKKC6L","created_at":"2026-06-12T01:09:46Z"},{"alias_kind":"pith_short_16","alias_value":"E47PO7DKKC6L4TXC","created_at":"2026-06-12T01:09:46Z"},{"alias_kind":"pith_short_8","alias_value":"E47PO7DK","created_at":"2026-06-12T01:09:46Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:E47PO7DKKC6L4TXC722UOYQWYD","target":"record","payload":{"canonical_record":{"source":{"id":"2606.13209","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-11T11:19:03Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"00793d84fef45565cf4a29d94d7fc2f91518d84b5e545caf5ddacdc29806f9d4","abstract_canon_sha256":"38a4aa43ee0c482cd8359e4cb2f19cc5baf3af1556bd16839fa03b30f8790762"},"schema_version":"1.0"},"canonical_sha256":"273ef77c6a50bcbe4ee2feb5476216c0d1c259843e4f140c3c02f580a299085d","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-12T01:09:46.870816Z","signature_b64":"roKX12qLxe54jtp8XMQQCA/bISs8xeEYSB+WTnFCpq1AuQV1xSXgcVEP/fUDypll6PEBji56ZNGzFjLsg8mUDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"273ef77c6a50bcbe4ee2feb5476216c0d1c259843e4f140c3c02f580a299085d","last_reissued_at":"2026-06-12T01:09:46.870180Z","signature_status":"signed_v1","first_computed_at":"2026-06-12T01:09:46.870180Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2606.13209","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":"2026-06-12T01:09:46Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"BTrj+ws1EsFX4/EOJ/6sovTXTTILZ8f1Eg+n2QamzGPb8J6GnWps89tNRqGcLV44cOFtP+cJeOlN9PQcmmg8AA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-22T12:47:57.925180Z"},"content_sha256":"ff4cced3256a2dac26464d46666393216527defbcc366674f07222110bb79c54","schema_version":"1.0","event_id":"sha256:ff4cced3256a2dac26464d46666393216527defbcc366674f07222110bb79c54"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:E47PO7DKKC6L4TXC722UOYQWYD","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Understanding helpfulness and harmless tension in reward models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Eshaan Tanwar, Pepa Atanasova","submitted_at":"2026-06-11T11:19:03Z","abstract_excerpt":"Reward models are a key component of reinforcement learning from human feedback (RLHF), aligning language models toward both helpful and harmless behaviour. However, the internal mechanisms underlying these objectives and their conflicts remain poorly understood. We study alignment tension in reward models trained under helpfulness-only, harmlessness-only, and mixed-objective settings. We find that mixed-objective models often underperform single-objective models, indicating interference between objectives. Using activation-based methods, we identify neurons associated with each objective and "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.13209","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/2606.13209/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":"2026-06-12T01:09:46Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"vu7FTr6QfnvcVkoByfxmHnF84w3Q0BwVDeQuFX4VKzrUwNPEufwTfInLjE0qdclTwYI6mtO15e8UHLVa5QDjDA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-22T12:47:57.925570Z"},"content_sha256":"d0723cafa46275ef2d41b85b5a2f8b5d34bca24efd29fb7c1927cf5703f661aa","schema_version":"1.0","event_id":"sha256:d0723cafa46275ef2d41b85b5a2f8b5d34bca24efd29fb7c1927cf5703f661aa"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/E47PO7DKKC6L4TXC722UOYQWYD/bundle.json","state_url":"https://pith.science/pith/E47PO7DKKC6L4TXC722UOYQWYD/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/E47PO7DKKC6L4TXC722UOYQWYD/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":"2026-06-22T12:47:57Z","links":{"resolver":"https://pith.science/pith/E47PO7DKKC6L4TXC722UOYQWYD","bundle":"https://pith.science/pith/E47PO7DKKC6L4TXC722UOYQWYD/bundle.json","state":"https://pith.science/pith/E47PO7DKKC6L4TXC722UOYQWYD/state.json","well_known_bundle":"https://pith.science/.well-known/pith/E47PO7DKKC6L4TXC722UOYQWYD/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:E47PO7DKKC6L4TXC722UOYQWYD","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":"38a4aa43ee0c482cd8359e4cb2f19cc5baf3af1556bd16839fa03b30f8790762","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-11T11:19:03Z","title_canon_sha256":"00793d84fef45565cf4a29d94d7fc2f91518d84b5e545caf5ddacdc29806f9d4"},"schema_version":"1.0","source":{"id":"2606.13209","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.13209","created_at":"2026-06-12T01:09:46Z"},{"alias_kind":"arxiv_version","alias_value":"2606.13209v1","created_at":"2026-06-12T01:09:46Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.13209","created_at":"2026-06-12T01:09:46Z"},{"alias_kind":"pith_short_12","alias_value":"E47PO7DKKC6L","created_at":"2026-06-12T01:09:46Z"},{"alias_kind":"pith_short_16","alias_value":"E47PO7DKKC6L4TXC","created_at":"2026-06-12T01:09:46Z"},{"alias_kind":"pith_short_8","alias_value":"E47PO7DK","created_at":"2026-06-12T01:09:46Z"}],"graph_snapshots":[{"event_id":"sha256:d0723cafa46275ef2d41b85b5a2f8b5d34bca24efd29fb7c1927cf5703f661aa","target":"graph","created_at":"2026-06-12T01:09:46Z","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/2606.13209/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Reward models are a key component of reinforcement learning from human feedback (RLHF), aligning language models toward both helpful and harmless behaviour. However, the internal mechanisms underlying these objectives and their conflicts remain poorly understood. We study alignment tension in reward models trained under helpfulness-only, harmlessness-only, and mixed-objective settings. We find that mixed-objective models often underperform single-objective models, indicating interference between objectives. Using activation-based methods, we identify neurons associated with each objective and ","authors_text":"Eshaan Tanwar, Pepa Atanasova","cross_cats":["cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-11T11:19:03Z","title":"Understanding helpfulness and harmless tension in reward models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.13209","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:ff4cced3256a2dac26464d46666393216527defbcc366674f07222110bb79c54","target":"record","created_at":"2026-06-12T01:09:46Z","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":"38a4aa43ee0c482cd8359e4cb2f19cc5baf3af1556bd16839fa03b30f8790762","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-11T11:19:03Z","title_canon_sha256":"00793d84fef45565cf4a29d94d7fc2f91518d84b5e545caf5ddacdc29806f9d4"},"schema_version":"1.0","source":{"id":"2606.13209","kind":"arxiv","version":1}},"canonical_sha256":"273ef77c6a50bcbe4ee2feb5476216c0d1c259843e4f140c3c02f580a299085d","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"273ef77c6a50bcbe4ee2feb5476216c0d1c259843e4f140c3c02f580a299085d","first_computed_at":"2026-06-12T01:09:46.870180Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-12T01:09:46.870180Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"roKX12qLxe54jtp8XMQQCA/bISs8xeEYSB+WTnFCpq1AuQV1xSXgcVEP/fUDypll6PEBji56ZNGzFjLsg8mUDQ==","signature_status":"signed_v1","signed_at":"2026-06-12T01:09:46.870816Z","signed_message":"canonical_sha256_bytes"},"source_id":"2606.13209","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:ff4cced3256a2dac26464d46666393216527defbcc366674f07222110bb79c54","sha256:d0723cafa46275ef2d41b85b5a2f8b5d34bca24efd29fb7c1927cf5703f661aa"],"state_sha256":"8a937c8a6d8b3f6ee31de4a82906d35a5ce6d2acf89d6b294c138859269aa04e"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Wx5+sloQaizNiAqgPv1e+PjDOy0v16HbDJ3czANX7ZTXw4QfqRO5RcJVoQFtoQOSbGyb6jSv+G/7jVvEX0qdBA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-22T12:47:57.927617Z","bundle_sha256":"6a79fa48c5094d0a72e68fa3325af263103e925131295c4d6ed254986203d2cd"}}