{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:IZTHVNI3BCMWL6Z7V6HBT5JKVG","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":"a3443d65e80c6b20daeb485eb7c94d218977e84eee4e94e1b2a145d1be403839","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-05T09:39:34Z","title_canon_sha256":"0f1a454297c8edc48a1235f8aeb13b16bc2f9c4241b67fc4339d495557f36e77"},"schema_version":"1.0","source":{"id":"2502.03032","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2502.03032","created_at":"2026-07-05T11:42:53Z"},{"alias_kind":"arxiv_version","alias_value":"2502.03032v3","created_at":"2026-07-05T11:42:53Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.03032","created_at":"2026-07-05T11:42:53Z"},{"alias_kind":"pith_short_12","alias_value":"IZTHVNI3BCMW","created_at":"2026-07-05T11:42:53Z"},{"alias_kind":"pith_short_16","alias_value":"IZTHVNI3BCMWL6Z7","created_at":"2026-07-05T11:42:53Z"},{"alias_kind":"pith_short_8","alias_value":"IZTHVNI3","created_at":"2026-07-05T11:42:53Z"}],"graph_snapshots":[{"event_id":"sha256:1a5b4db9ef7b3d33d2ed92874e1feebe3f7d4fb7231e7432b633129e99371a33","target":"graph","created_at":"2026-07-05T11:42:53Z","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/2502.03032/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"We introduce a new approach to systematically map features discovered by sparse autoencoder across consecutive layers of large language models, extending earlier work that examined inter-layer feature links. By using a data-free cosine similarity technique, we trace how specific features persist, transform, or first appear at each stage. This method yields granular flow graphs of feature evolution, enabling fine-grained interpretability and mechanistic insights into model computations. Crucially, we demonstrate how these cross-layer feature maps facilitate direct steering of model behavior by ","authors_text":"Daniil Gavrilov, Daniil Laptev, Nikita Balagansky, Yaroslav Aksenov","cross_cats":["cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-05T09:39:34Z","title":"Analyze Feature Flow to Enhance Interpretation and Steering in Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.03032","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":"sha256:5fcabc881487b2eddce97ff8ca8c91b52d4df6ffb590799c010f0965f8049ccc","target":"record","created_at":"2026-07-05T11:42:53Z","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":"a3443d65e80c6b20daeb485eb7c94d218977e84eee4e94e1b2a145d1be403839","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-05T09:39:34Z","title_canon_sha256":"0f1a454297c8edc48a1235f8aeb13b16bc2f9c4241b67fc4339d495557f36e77"},"schema_version":"1.0","source":{"id":"2502.03032","kind":"arxiv","version":3}},"canonical_sha256":"46667ab51b089965fb3faf8e19f52aa9a51666562502c94c763de0cdf88e66ba","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"46667ab51b089965fb3faf8e19f52aa9a51666562502c94c763de0cdf88e66ba","first_computed_at":"2026-07-05T11:42:53.256107Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:42:53.256107Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"PZZj0sNQw5VkatYmupqdDDJo5Wv5DzPxbp4OzrAB4n31W1g/aJ6jNTSk3cbzhcpShHNBZYWfYsH9p+wQQFPVAg==","signature_status":"signed_v1","signed_at":"2026-07-05T11:42:53.256609Z","signed_message":"canonical_sha256_bytes"},"source_id":"2502.03032","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:5fcabc881487b2eddce97ff8ca8c91b52d4df6ffb590799c010f0965f8049ccc","sha256:1a5b4db9ef7b3d33d2ed92874e1feebe3f7d4fb7231e7432b633129e99371a33"],"state_sha256":"67bd858591da5f1cea61e7b62f2d481382f0092078afd2bd7f809c70323c194f"}