{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:VFW436TZ6XUKQPWF47MXUGDNQB","short_pith_number":"pith:VFW436TZ","canonical_record":{"source":{"id":"2506.05166","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-06-05T15:43:34Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"31c75b983ef7f237c3b05a11238b1bafbd6387ada7acd56b58ea1330bd63cc65","abstract_canon_sha256":"618376083df4e72b7fa1a5ab05753829074422dde6dce2d52057a7a219795439"},"schema_version":"1.0"},"canonical_sha256":"a96dcdfa79f5e8a83ec5e7d97a186d807cb9e9af614f71f4ce9bdcd61968bf6c","source":{"kind":"arxiv","id":"2506.05166","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2506.05166","created_at":"2026-07-05T11:17:00Z"},{"alias_kind":"arxiv_version","alias_value":"2506.05166v2","created_at":"2026-07-05T11:17:00Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.05166","created_at":"2026-07-05T11:17:00Z"},{"alias_kind":"pith_short_12","alias_value":"VFW436TZ6XUK","created_at":"2026-07-05T11:17:00Z"},{"alias_kind":"pith_short_16","alias_value":"VFW436TZ6XUKQPWF","created_at":"2026-07-05T11:17:00Z"},{"alias_kind":"pith_short_8","alias_value":"VFW436TZ","created_at":"2026-07-05T11:17:00Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:VFW436TZ6XUKQPWF47MXUGDNQB","target":"record","payload":{"canonical_record":{"source":{"id":"2506.05166","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-06-05T15:43:34Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"31c75b983ef7f237c3b05a11238b1bafbd6387ada7acd56b58ea1330bd63cc65","abstract_canon_sha256":"618376083df4e72b7fa1a5ab05753829074422dde6dce2d52057a7a219795439"},"schema_version":"1.0"},"canonical_sha256":"a96dcdfa79f5e8a83ec5e7d97a186d807cb9e9af614f71f4ce9bdcd61968bf6c","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:17:00.824613Z","signature_b64":"eElCdFx5oSv02G+FRVv+0K79o73uqWgekYdBZj87vJHqrsOUtLGE2IoWCcBSfUA5JoJCCmLptC3s/yY6jVqUCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a96dcdfa79f5e8a83ec5e7d97a186d807cb9e9af614f71f4ce9bdcd61968bf6c","last_reissued_at":"2026-07-05T11:17:00.824090Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:17:00.824090Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2506.05166","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:17:00Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"5LyevgjiyCa/t530oN3fEmLqRQYA4TMGNcjFjDV/sQkkFWKChhK4ToRLKY5lwCFX3+n9n4b8xql3uIdr42EaAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T08:51:12.959970Z"},"content_sha256":"c68db7d4efe8dc235f97a7ea03526650c225dee1b1b7b459d03de3edda561ba9","schema_version":"1.0","event_id":"sha256:c68db7d4efe8dc235f97a7ea03526650c225dee1b1b7b459d03de3edda561ba9"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:VFW436TZ6XUKQPWF47MXUGDNQB","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Dissecting Bias in LLMs: A Mechanistic Interpretability Perspective","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Bhavik Chandna, Procheta Sen, Zubair Bashir","submitted_at":"2025-06-05T15:43:34Z","abstract_excerpt":"Large Language Models (LLMs) are known to exhibit social, demographic, and gender biases, often as a consequence of the data on which they are trained. In this work, we adopt a mechanistic interpretability approach to analyze how such biases are structurally represented within models such as GPT-2 and Llama2. Focusing on demographic and gender biases, we explore different metrics to identify the internal edges responsible for biased behavior. We then assess the stability, localization, and generalizability of these components across dataset and linguistic variations. Through systematic ablatio"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.05166","kind":"arxiv","version":2},"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/2506.05166/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":"2026-07-05T11:17:00Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"x0J0OS9pTkSRzb5S/z2a0gu0RjWFwu5aVJ7GdHmEU+KZI3T/AQXxXhGAbTsznF/NihZbdpvb9rJLI1KFE4WSAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T08:51:12.960893Z"},"content_sha256":"cbcb157bace99d976364d7a380c8ae052f9c659ae1a9ca9047b4c147f2ec7e73","schema_version":"1.0","event_id":"sha256:cbcb157bace99d976364d7a380c8ae052f9c659ae1a9ca9047b4c147f2ec7e73"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/VFW436TZ6XUKQPWF47MXUGDNQB/bundle.json","state_url":"https://pith.science/pith/VFW436TZ6XUKQPWF47MXUGDNQB/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/VFW436TZ6XUKQPWF47MXUGDNQB/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":"2026-08-09T08:51:12Z","links":{"resolver":"https://pith.science/pith/VFW436TZ6XUKQPWF47MXUGDNQB","bundle":"https://pith.science/pith/VFW436TZ6XUKQPWF47MXUGDNQB/bundle.json","state":"https://pith.science/pith/VFW436TZ6XUKQPWF47MXUGDNQB/state.json","well_known_bundle":"https://pith.science/.well-known/pith/VFW436TZ6XUKQPWF47MXUGDNQB/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:VFW436TZ6XUKQPWF47MXUGDNQB","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":"618376083df4e72b7fa1a5ab05753829074422dde6dce2d52057a7a219795439","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-06-05T15:43:34Z","title_canon_sha256":"31c75b983ef7f237c3b05a11238b1bafbd6387ada7acd56b58ea1330bd63cc65"},"schema_version":"1.0","source":{"id":"2506.05166","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2506.05166","created_at":"2026-07-05T11:17:00Z"},{"alias_kind":"arxiv_version","alias_value":"2506.05166v2","created_at":"2026-07-05T11:17:00Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.05166","created_at":"2026-07-05T11:17:00Z"},{"alias_kind":"pith_short_12","alias_value":"VFW436TZ6XUK","created_at":"2026-07-05T11:17:00Z"},{"alias_kind":"pith_short_16","alias_value":"VFW436TZ6XUKQPWF","created_at":"2026-07-05T11:17:00Z"},{"alias_kind":"pith_short_8","alias_value":"VFW436TZ","created_at":"2026-07-05T11:17:00Z"}],"graph_snapshots":[{"event_id":"sha256:cbcb157bace99d976364d7a380c8ae052f9c659ae1a9ca9047b4c147f2ec7e73","target":"graph","created_at":"2026-07-05T11:17:00Z","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/2506.05166/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Large Language Models (LLMs) are known to exhibit social, demographic, and gender biases, often as a consequence of the data on which they are trained. In this work, we adopt a mechanistic interpretability approach to analyze how such biases are structurally represented within models such as GPT-2 and Llama2. Focusing on demographic and gender biases, we explore different metrics to identify the internal edges responsible for biased behavior. We then assess the stability, localization, and generalizability of these components across dataset and linguistic variations. Through systematic ablatio","authors_text":"Bhavik Chandna, Procheta Sen, Zubair Bashir","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-06-05T15:43:34Z","title":"Dissecting Bias in LLMs: A Mechanistic Interpretability Perspective"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.05166","kind":"arxiv","version":2},"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:c68db7d4efe8dc235f97a7ea03526650c225dee1b1b7b459d03de3edda561ba9","target":"record","created_at":"2026-07-05T11:17:00Z","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":"618376083df4e72b7fa1a5ab05753829074422dde6dce2d52057a7a219795439","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-06-05T15:43:34Z","title_canon_sha256":"31c75b983ef7f237c3b05a11238b1bafbd6387ada7acd56b58ea1330bd63cc65"},"schema_version":"1.0","source":{"id":"2506.05166","kind":"arxiv","version":2}},"canonical_sha256":"a96dcdfa79f5e8a83ec5e7d97a186d807cb9e9af614f71f4ce9bdcd61968bf6c","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"a96dcdfa79f5e8a83ec5e7d97a186d807cb9e9af614f71f4ce9bdcd61968bf6c","first_computed_at":"2026-07-05T11:17:00.824090Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:17:00.824090Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"eElCdFx5oSv02G+FRVv+0K79o73uqWgekYdBZj87vJHqrsOUtLGE2IoWCcBSfUA5JoJCCmLptC3s/yY6jVqUCg==","signature_status":"signed_v1","signed_at":"2026-07-05T11:17:00.824613Z","signed_message":"canonical_sha256_bytes"},"source_id":"2506.05166","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c68db7d4efe8dc235f97a7ea03526650c225dee1b1b7b459d03de3edda561ba9","sha256:cbcb157bace99d976364d7a380c8ae052f9c659ae1a9ca9047b4c147f2ec7e73"],"state_sha256":"f53973459150e86460a2abb25b16c5810b15b67ea19ad6084f725038e82ce65c"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"CPkPm04IiUMvFNKTKCYU93K258/zu6+pcoW/4XqlG2q0uItkhxyUZiBpGiS6+Z/YeYXWtkknXQZKnQPIFN2/AQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-09T08:51:12.977187Z","bundle_sha256":"1284607d537bb18417454318edcb6d0835cb873dff4cd8321d01a88d1ab16825"}}