{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:FWW45B65N22SQJTGMQ4DXMPYIF","short_pith_number":"pith:FWW45B65","canonical_record":{"source":{"id":"2504.04215","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-05T16:00:44Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"58dbda73e7d18fa8fd8848c8125751498867706bc247f5c8c98d3d986c69ebcc","abstract_canon_sha256":"87c3a1bea83090e33907a729d92f23904245b1ec58d3f43b58c78260a6ddad4b"},"schema_version":"1.0"},"canonical_sha256":"2dadce87dd6eb528266664383bb1f841480713d1d3a15a5a7fe410387287d170","source":{"kind":"arxiv","id":"2504.04215","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2504.04215","created_at":"2026-07-05T10:44:55Z"},{"alias_kind":"arxiv_version","alias_value":"2504.04215v1","created_at":"2026-07-05T10:44:55Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.04215","created_at":"2026-07-05T10:44:55Z"},{"alias_kind":"pith_short_12","alias_value":"FWW45B65N22S","created_at":"2026-07-05T10:44:55Z"},{"alias_kind":"pith_short_16","alias_value":"FWW45B65N22SQJTG","created_at":"2026-07-05T10:44:55Z"},{"alias_kind":"pith_short_8","alias_value":"FWW45B65","created_at":"2026-07-05T10:44:55Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:FWW45B65N22SQJTGMQ4DXMPYIF","target":"record","payload":{"canonical_record":{"source":{"id":"2504.04215","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-05T16:00:44Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"58dbda73e7d18fa8fd8848c8125751498867706bc247f5c8c98d3d986c69ebcc","abstract_canon_sha256":"87c3a1bea83090e33907a729d92f23904245b1ec58d3f43b58c78260a6ddad4b"},"schema_version":"1.0"},"canonical_sha256":"2dadce87dd6eb528266664383bb1f841480713d1d3a15a5a7fe410387287d170","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:44:55.885056Z","signature_b64":"0Q9iIhSdcnKQLUIKu8lA7vW/hchO/krd5o+FOUHvBWIbggiTGpFqnVH09R+7DQEG76P1DgTy/509I+iP8GwuCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2dadce87dd6eb528266664383bb1f841480713d1d3a15a5a7fe410387287d170","last_reissued_at":"2026-07-05T10:44:55.884533Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:44:55.884533Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2504.04215","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":"2026-07-05T10:44:55Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"w4/j68RMt+oJeDHrE6jPEl4DcGwDcY3xKZQ+l3EcISxX0Vj9JHK7PCQ/xfOT9mJB4uA5fzJMaQD5GebgrnyMAQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-11T02:10:27.891306Z"},"content_sha256":"4bbc18647330fd50e0b47bfe0edd0ac8aed5bb5118f2bd71965bf393044fc974","schema_version":"1.0","event_id":"sha256:4bbc18647330fd50e0b47bfe0edd0ac8aed5bb5118f2bd71965bf393044fc974"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:FWW45B65N22SQJTGMQ4DXMPYIF","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Towards Understanding and Improving Refusal in Compressed Models via Mechanistic Interpretability","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Mohammad Mahdi Khalili, Vishnu Kabir Chhabra","submitted_at":"2025-04-05T16:00:44Z","abstract_excerpt":"The rapid growth of large language models has spurred significant interest in model compression as a means to enhance their accessibility and practicality. While extensive research has explored model compression through the lens of safety, findings suggest that safety-aligned models often lose elements of trustworthiness post-compression. Simultaneously, the field of mechanistic interpretability has gained traction, with notable discoveries, such as the identification of a single direction in the residual stream mediating refusal behaviors across diverse model architectures. In this work, we i"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.04215","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/2504.04215/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":"2026-07-05T10:44:55Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"pI0fIiV4CjI0D4qMOGQ0ir0fcvo7HSa9Qnu9YPv8gDeVvYcQyyR8N/PlyVwQUJWuzrFEJKvzwmOBJhdm6TNdAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-11T02:10:27.892253Z"},"content_sha256":"a550cb5d86f586ba97e723d392261def533e71534aa484b40d7ad51b138c0dfe","schema_version":"1.0","event_id":"sha256:a550cb5d86f586ba97e723d392261def533e71534aa484b40d7ad51b138c0dfe"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/FWW45B65N22SQJTGMQ4DXMPYIF/bundle.json","state_url":"https://pith.science/pith/FWW45B65N22SQJTGMQ4DXMPYIF/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/FWW45B65N22SQJTGMQ4DXMPYIF/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":"2026-08-11T02:10:27Z","links":{"resolver":"https://pith.science/pith/FWW45B65N22SQJTGMQ4DXMPYIF","bundle":"https://pith.science/pith/FWW45B65N22SQJTGMQ4DXMPYIF/bundle.json","state":"https://pith.science/pith/FWW45B65N22SQJTGMQ4DXMPYIF/state.json","well_known_bundle":"https://pith.science/.well-known/pith/FWW45B65N22SQJTGMQ4DXMPYIF/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:FWW45B65N22SQJTGMQ4DXMPYIF","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":"87c3a1bea83090e33907a729d92f23904245b1ec58d3f43b58c78260a6ddad4b","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-05T16:00:44Z","title_canon_sha256":"58dbda73e7d18fa8fd8848c8125751498867706bc247f5c8c98d3d986c69ebcc"},"schema_version":"1.0","source":{"id":"2504.04215","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2504.04215","created_at":"2026-07-05T10:44:55Z"},{"alias_kind":"arxiv_version","alias_value":"2504.04215v1","created_at":"2026-07-05T10:44:55Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.04215","created_at":"2026-07-05T10:44:55Z"},{"alias_kind":"pith_short_12","alias_value":"FWW45B65N22S","created_at":"2026-07-05T10:44:55Z"},{"alias_kind":"pith_short_16","alias_value":"FWW45B65N22SQJTG","created_at":"2026-07-05T10:44:55Z"},{"alias_kind":"pith_short_8","alias_value":"FWW45B65","created_at":"2026-07-05T10:44:55Z"}],"graph_snapshots":[{"event_id":"sha256:a550cb5d86f586ba97e723d392261def533e71534aa484b40d7ad51b138c0dfe","target":"graph","created_at":"2026-07-05T10:44:55Z","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/2504.04215/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"The rapid growth of large language models has spurred significant interest in model compression as a means to enhance their accessibility and practicality. While extensive research has explored model compression through the lens of safety, findings suggest that safety-aligned models often lose elements of trustworthiness post-compression. Simultaneously, the field of mechanistic interpretability has gained traction, with notable discoveries, such as the identification of a single direction in the residual stream mediating refusal behaviors across diverse model architectures. In this work, we i","authors_text":"Mohammad Mahdi Khalili, Vishnu Kabir Chhabra","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-05T16:00:44Z","title":"Towards Understanding and Improving Refusal in Compressed Models via Mechanistic Interpretability"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.04215","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":"sha256:4bbc18647330fd50e0b47bfe0edd0ac8aed5bb5118f2bd71965bf393044fc974","target":"record","created_at":"2026-07-05T10:44:55Z","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":"87c3a1bea83090e33907a729d92f23904245b1ec58d3f43b58c78260a6ddad4b","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-05T16:00:44Z","title_canon_sha256":"58dbda73e7d18fa8fd8848c8125751498867706bc247f5c8c98d3d986c69ebcc"},"schema_version":"1.0","source":{"id":"2504.04215","kind":"arxiv","version":1}},"canonical_sha256":"2dadce87dd6eb528266664383bb1f841480713d1d3a15a5a7fe410387287d170","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"2dadce87dd6eb528266664383bb1f841480713d1d3a15a5a7fe410387287d170","first_computed_at":"2026-07-05T10:44:55.884533Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:44:55.884533Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"0Q9iIhSdcnKQLUIKu8lA7vW/hchO/krd5o+FOUHvBWIbggiTGpFqnVH09R+7DQEG76P1DgTy/509I+iP8GwuCA==","signature_status":"signed_v1","signed_at":"2026-07-05T10:44:55.885056Z","signed_message":"canonical_sha256_bytes"},"source_id":"2504.04215","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:4bbc18647330fd50e0b47bfe0edd0ac8aed5bb5118f2bd71965bf393044fc974","sha256:a550cb5d86f586ba97e723d392261def533e71534aa484b40d7ad51b138c0dfe"],"state_sha256":"27ee9d0f4eb3d086d27033b9cb8b61ac263990310a61ea43154dc5643fe9776a"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"tu3+J9HOVMGD7KsGOINi6+kLNCGK2zQBZwC9RT0CRbEiirRe7DgYtWBJH9kaDTE23Nug7c4pYHC+SBrkdrOACg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-11T02:10:27.899531Z","bundle_sha256":"30d73a01d9fe386e2df3d356b27799f2f9b07ce3b7f5b0a45885fac1f6518218"}}