{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:SIXU6TAN5V7K7UTXL5YJ4JIK5K","short_pith_number":"pith:SIXU6TAN","canonical_record":{"source":{"id":"2503.00177","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-28T20:43:45Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"8db1deb962c1a72d279263c8ac738a30239198d1f86a713fa2c1db4f9f322ee3","abstract_canon_sha256":"2b4e067f5ace942c0bcb0a8aef33113c9528073136aa3eb39a58ed3bea01fc09"},"schema_version":"1.0"},"canonical_sha256":"922f4f4c0ded7eafd2775f709e250aeaad14225a5fe0def7347ac914891d26a7","source":{"kind":"arxiv","id":"2503.00177","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2503.00177","created_at":"2026-07-05T10:22:18Z"},{"alias_kind":"arxiv_version","alias_value":"2503.00177v1","created_at":"2026-07-05T10:22:18Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.00177","created_at":"2026-07-05T10:22:18Z"},{"alias_kind":"pith_short_12","alias_value":"SIXU6TAN5V7K","created_at":"2026-07-05T10:22:18Z"},{"alias_kind":"pith_short_16","alias_value":"SIXU6TAN5V7K7UTX","created_at":"2026-07-05T10:22:18Z"},{"alias_kind":"pith_short_8","alias_value":"SIXU6TAN","created_at":"2026-07-05T10:22:18Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:SIXU6TAN5V7K7UTXL5YJ4JIK5K","target":"record","payload":{"canonical_record":{"source":{"id":"2503.00177","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-28T20:43:45Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"8db1deb962c1a72d279263c8ac738a30239198d1f86a713fa2c1db4f9f322ee3","abstract_canon_sha256":"2b4e067f5ace942c0bcb0a8aef33113c9528073136aa3eb39a58ed3bea01fc09"},"schema_version":"1.0"},"canonical_sha256":"922f4f4c0ded7eafd2775f709e250aeaad14225a5fe0def7347ac914891d26a7","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:22:18.605998Z","signature_b64":"dclsbTG9BNKBN9CcyC265FYECDX58uviOtrTOMOdBzgS6EK8L5dQrvFVso1tdLhArepHOrvGYUWkgkTkmyqyDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"922f4f4c0ded7eafd2775f709e250aeaad14225a5fe0def7347ac914891d26a7","last_reissued_at":"2026-07-05T10:22:18.605451Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:22:18.605451Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2503.00177","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":"2026-07-05T10:22:18Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"G5ruj1WWm5Vm/wCOEACGtWv883E0PcapKkWOCSyvFZ/GhfzMee3RkTaLJZKuXAhMiuqUifPzbIVagV+NDDeVDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T03:09:50.360201Z"},"content_sha256":"8698e490ded3b42f2abe000b3a1b911e9a2a5e5a96b69fdba571a0e32be18e4f","schema_version":"1.0","event_id":"sha256:8698e490ded3b42f2abe000b3a1b911e9a2a5e5a96b69fdba571a0e32be18e4f"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:SIXU6TAN5V7K7UTXL5YJ4JIK5K","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Steering Large Language Model Activations in Sparse Spaces","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Ali Rahimi-Kalahroudi, Mohammad Pezeshki, Pascal Vincent, Reza Bayat, Sarath Chandar","submitted_at":"2025-02-28T20:43:45Z","abstract_excerpt":"A key challenge in AI alignment is guiding large language models (LLMs) to follow desired behaviors at test time. Activation steering, which modifies internal model activations during inference, offers a potential solution. However, prior work in dense activation spaces struggles with superposition, wherein multiple features become entangled, limiting interpretability and precise control. In contrast, sparse representations provide an untapped opportunity for more interpretable behavior modulation. In this work, we introduce sparse activation steering (SAS), a method that leverages sparse auto"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.00177","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/2503.00177/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":"2026-07-05T10:22:18Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"lJ95PaylvUbZmdwI3gVvvCfMbQ4mM/g4FwNer9Fq6QscGo0xyVvWyiwiS/0WSRneXnl2nidBON9zuZEVGDDRDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T03:09:50.361121Z"},"content_sha256":"46e8dad784fad42c4db6a5e682e93a5d7703c6f2a3d02cef57a6040617f9e4f4","schema_version":"1.0","event_id":"sha256:46e8dad784fad42c4db6a5e682e93a5d7703c6f2a3d02cef57a6040617f9e4f4"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/SIXU6TAN5V7K7UTXL5YJ4JIK5K/bundle.json","state_url":"https://pith.science/pith/SIXU6TAN5V7K7UTXL5YJ4JIK5K/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/SIXU6TAN5V7K7UTXL5YJ4JIK5K/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":"2026-08-06T03:09:50Z","links":{"resolver":"https://pith.science/pith/SIXU6TAN5V7K7UTXL5YJ4JIK5K","bundle":"https://pith.science/pith/SIXU6TAN5V7K7UTXL5YJ4JIK5K/bundle.json","state":"https://pith.science/pith/SIXU6TAN5V7K7UTXL5YJ4JIK5K/state.json","well_known_bundle":"https://pith.science/.well-known/pith/SIXU6TAN5V7K7UTXL5YJ4JIK5K/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:SIXU6TAN5V7K7UTXL5YJ4JIK5K","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":"2b4e067f5ace942c0bcb0a8aef33113c9528073136aa3eb39a58ed3bea01fc09","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-28T20:43:45Z","title_canon_sha256":"8db1deb962c1a72d279263c8ac738a30239198d1f86a713fa2c1db4f9f322ee3"},"schema_version":"1.0","source":{"id":"2503.00177","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2503.00177","created_at":"2026-07-05T10:22:18Z"},{"alias_kind":"arxiv_version","alias_value":"2503.00177v1","created_at":"2026-07-05T10:22:18Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.00177","created_at":"2026-07-05T10:22:18Z"},{"alias_kind":"pith_short_12","alias_value":"SIXU6TAN5V7K","created_at":"2026-07-05T10:22:18Z"},{"alias_kind":"pith_short_16","alias_value":"SIXU6TAN5V7K7UTX","created_at":"2026-07-05T10:22:18Z"},{"alias_kind":"pith_short_8","alias_value":"SIXU6TAN","created_at":"2026-07-05T10:22:18Z"}],"graph_snapshots":[{"event_id":"sha256:46e8dad784fad42c4db6a5e682e93a5d7703c6f2a3d02cef57a6040617f9e4f4","target":"graph","created_at":"2026-07-05T10:22:18Z","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/2503.00177/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"A key challenge in AI alignment is guiding large language models (LLMs) to follow desired behaviors at test time. Activation steering, which modifies internal model activations during inference, offers a potential solution. However, prior work in dense activation spaces struggles with superposition, wherein multiple features become entangled, limiting interpretability and precise control. In contrast, sparse representations provide an untapped opportunity for more interpretable behavior modulation. In this work, we introduce sparse activation steering (SAS), a method that leverages sparse auto","authors_text":"Ali Rahimi-Kalahroudi, Mohammad Pezeshki, Pascal Vincent, Reza Bayat, Sarath Chandar","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-28T20:43:45Z","title":"Steering Large Language Model Activations in Sparse Spaces"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.00177","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":"sha256:8698e490ded3b42f2abe000b3a1b911e9a2a5e5a96b69fdba571a0e32be18e4f","target":"record","created_at":"2026-07-05T10:22:18Z","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":"2b4e067f5ace942c0bcb0a8aef33113c9528073136aa3eb39a58ed3bea01fc09","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-28T20:43:45Z","title_canon_sha256":"8db1deb962c1a72d279263c8ac738a30239198d1f86a713fa2c1db4f9f322ee3"},"schema_version":"1.0","source":{"id":"2503.00177","kind":"arxiv","version":1}},"canonical_sha256":"922f4f4c0ded7eafd2775f709e250aeaad14225a5fe0def7347ac914891d26a7","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"922f4f4c0ded7eafd2775f709e250aeaad14225a5fe0def7347ac914891d26a7","first_computed_at":"2026-07-05T10:22:18.605451Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:22:18.605451Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"dclsbTG9BNKBN9CcyC265FYECDX58uviOtrTOMOdBzgS6EK8L5dQrvFVso1tdLhArepHOrvGYUWkgkTkmyqyDA==","signature_status":"signed_v1","signed_at":"2026-07-05T10:22:18.605998Z","signed_message":"canonical_sha256_bytes"},"source_id":"2503.00177","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:8698e490ded3b42f2abe000b3a1b911e9a2a5e5a96b69fdba571a0e32be18e4f","sha256:46e8dad784fad42c4db6a5e682e93a5d7703c6f2a3d02cef57a6040617f9e4f4"],"state_sha256":"e0680adab6cb227a818815ffc2856b09c3a204abf4a1a7aa6155e09dc1c8e084"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"el54u3xfuNtsebFRn/OcR5a/yMKXBqXUP29yDuXf+h4kAfMSrXA0l50PKr75rvq1p2iQoIzNraYmrRo9HMKyAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T03:09:50.410840Z","bundle_sha256":"81af9502e1b70936459f6d38328b4fb1f4ae5f1a067ddb4be37aa1497981c2ba"}}