{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:6KJVPNXMOCIAAIAUUNXNP5KEES","short_pith_number":"pith:6KJVPNXM","canonical_record":{"source":{"id":"2312.12736","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-12-20T03:18:50Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"0d61a7a9808535e0e338d9b9e854f0c3f5fe1820353f77a646c7ca14c161e5b4","abstract_canon_sha256":"c650fd9ee1f8b6e2ede7efa088cde2c368781b11610d2c9fa5751112aa498839"},"schema_version":"1.0"},"canonical_sha256":"f29357b6ec7090002014a36ed7f54424bd5bb7cd3cea1845d10583d203ec88e5","source":{"kind":"arxiv","id":"2312.12736","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2312.12736","created_at":"2026-07-05T08:39:29Z"},{"alias_kind":"arxiv_version","alias_value":"2312.12736v2","created_at":"2026-07-05T08:39:29Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2312.12736","created_at":"2026-07-05T08:39:29Z"},{"alias_kind":"pith_short_12","alias_value":"6KJVPNXMOCIA","created_at":"2026-07-05T08:39:29Z"},{"alias_kind":"pith_short_16","alias_value":"6KJVPNXMOCIAAIAU","created_at":"2026-07-05T08:39:29Z"},{"alias_kind":"pith_short_8","alias_value":"6KJVPNXM","created_at":"2026-07-05T08:39:29Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:6KJVPNXMOCIAAIAUUNXNP5KEES","target":"record","payload":{"canonical_record":{"source":{"id":"2312.12736","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-12-20T03:18:50Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"0d61a7a9808535e0e338d9b9e854f0c3f5fe1820353f77a646c7ca14c161e5b4","abstract_canon_sha256":"c650fd9ee1f8b6e2ede7efa088cde2c368781b11610d2c9fa5751112aa498839"},"schema_version":"1.0"},"canonical_sha256":"f29357b6ec7090002014a36ed7f54424bd5bb7cd3cea1845d10583d203ec88e5","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:39:29.936488Z","signature_b64":"dS4+fc3s3ok8q02X5ssfWdScUTikP5EY5Ki8D7D7e2j6NydCHD0ODS7XcHDO56WHEYPvQeTzrhG+Y8L4PrihBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f29357b6ec7090002014a36ed7f54424bd5bb7cd3cea1845d10583d203ec88e5","last_reissued_at":"2026-07-05T08:39:29.936020Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:39:29.936020Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2312.12736","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":"2026-07-05T08:39:29Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"1Rdi5XTljSu9NqdF/Da8GtRYzGjnV2G1t5a5/14pC81SOJv/MT9RbB+ZzK55aLVtY1lC0KTd6T1nmKp/e3ZICw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-12T08:54:26.808079Z"},"content_sha256":"e5dc0cc45c47fc93ad89830157c91ddb22a03d46fe41d96c456fb4de06fd0262","schema_version":"1.0","event_id":"sha256:e5dc0cc45c47fc93ad89830157c91ddb22a03d46fe41d96c456fb4de06fd0262"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:6KJVPNXMOCIAAIAUUNXNP5KEES","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Learning and Forgetting Unsafe Examples in Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"David Madras, James Zou, Jiachen Zhao, Mengye Ren, Zhun Deng","submitted_at":"2023-12-20T03:18:50Z","abstract_excerpt":"As the number of large language models (LLMs) released to the public grows, there is a pressing need to understand the safety implications associated with these models learning from third-party custom finetuning data. We explore the behavior of LLMs finetuned on noisy custom data containing unsafe content, represented by datasets that contain biases, toxicity, and harmfulness, finding that while aligned LLMs can readily learn this unsafe content, they also tend to forget it more significantly than other examples when subsequently finetuned on safer content. Drawing inspiration from the discrep"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2312.12736","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/2312.12736/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":"2026-07-05T08:39:29Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"ydPGyrs8xHr/3kwSQSfbljbESVKWn4b6KY1ztJj9PDrZVct2nBc6qsUmN99svazSnsvi6yW9KJGPL9KG4yCWCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-12T08:54:26.808937Z"},"content_sha256":"67340595b0d6e147d274a7ddf3b12d0d5c70b2c4ceaa19618919c09f32662906","schema_version":"1.0","event_id":"sha256:67340595b0d6e147d274a7ddf3b12d0d5c70b2c4ceaa19618919c09f32662906"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/6KJVPNXMOCIAAIAUUNXNP5KEES/bundle.json","state_url":"https://pith.science/pith/6KJVPNXMOCIAAIAUUNXNP5KEES/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/6KJVPNXMOCIAAIAUUNXNP5KEES/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":"2026-08-12T08:54:26Z","links":{"resolver":"https://pith.science/pith/6KJVPNXMOCIAAIAUUNXNP5KEES","bundle":"https://pith.science/pith/6KJVPNXMOCIAAIAUUNXNP5KEES/bundle.json","state":"https://pith.science/pith/6KJVPNXMOCIAAIAUUNXNP5KEES/state.json","well_known_bundle":"https://pith.science/.well-known/pith/6KJVPNXMOCIAAIAUUNXNP5KEES/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:6KJVPNXMOCIAAIAUUNXNP5KEES","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":"c650fd9ee1f8b6e2ede7efa088cde2c368781b11610d2c9fa5751112aa498839","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-12-20T03:18:50Z","title_canon_sha256":"0d61a7a9808535e0e338d9b9e854f0c3f5fe1820353f77a646c7ca14c161e5b4"},"schema_version":"1.0","source":{"id":"2312.12736","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2312.12736","created_at":"2026-07-05T08:39:29Z"},{"alias_kind":"arxiv_version","alias_value":"2312.12736v2","created_at":"2026-07-05T08:39:29Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2312.12736","created_at":"2026-07-05T08:39:29Z"},{"alias_kind":"pith_short_12","alias_value":"6KJVPNXMOCIA","created_at":"2026-07-05T08:39:29Z"},{"alias_kind":"pith_short_16","alias_value":"6KJVPNXMOCIAAIAU","created_at":"2026-07-05T08:39:29Z"},{"alias_kind":"pith_short_8","alias_value":"6KJVPNXM","created_at":"2026-07-05T08:39:29Z"}],"graph_snapshots":[{"event_id":"sha256:67340595b0d6e147d274a7ddf3b12d0d5c70b2c4ceaa19618919c09f32662906","target":"graph","created_at":"2026-07-05T08:39:29Z","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/2312.12736/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"As the number of large language models (LLMs) released to the public grows, there is a pressing need to understand the safety implications associated with these models learning from third-party custom finetuning data. We explore the behavior of LLMs finetuned on noisy custom data containing unsafe content, represented by datasets that contain biases, toxicity, and harmfulness, finding that while aligned LLMs can readily learn this unsafe content, they also tend to forget it more significantly than other examples when subsequently finetuned on safer content. Drawing inspiration from the discrep","authors_text":"David Madras, James Zou, Jiachen Zhao, Mengye Ren, Zhun Deng","cross_cats":["cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-12-20T03:18:50Z","title":"Learning and Forgetting Unsafe Examples in Large Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2312.12736","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:e5dc0cc45c47fc93ad89830157c91ddb22a03d46fe41d96c456fb4de06fd0262","target":"record","created_at":"2026-07-05T08:39:29Z","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":"c650fd9ee1f8b6e2ede7efa088cde2c368781b11610d2c9fa5751112aa498839","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2023-12-20T03:18:50Z","title_canon_sha256":"0d61a7a9808535e0e338d9b9e854f0c3f5fe1820353f77a646c7ca14c161e5b4"},"schema_version":"1.0","source":{"id":"2312.12736","kind":"arxiv","version":2}},"canonical_sha256":"f29357b6ec7090002014a36ed7f54424bd5bb7cd3cea1845d10583d203ec88e5","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"f29357b6ec7090002014a36ed7f54424bd5bb7cd3cea1845d10583d203ec88e5","first_computed_at":"2026-07-05T08:39:29.936020Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:39:29.936020Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"dS4+fc3s3ok8q02X5ssfWdScUTikP5EY5Ki8D7D7e2j6NydCHD0ODS7XcHDO56WHEYPvQeTzrhG+Y8L4PrihBA==","signature_status":"signed_v1","signed_at":"2026-07-05T08:39:29.936488Z","signed_message":"canonical_sha256_bytes"},"source_id":"2312.12736","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:e5dc0cc45c47fc93ad89830157c91ddb22a03d46fe41d96c456fb4de06fd0262","sha256:67340595b0d6e147d274a7ddf3b12d0d5c70b2c4ceaa19618919c09f32662906"],"state_sha256":"4602322ed482b82fe84b29ff8c44027d6a90734b623ed78900160056a38a4287"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"/yRKiNpSGUf+49GRG3gaGUAowhU9M+BF7BCP4qEvbq+urmm2ynbVpMCdc/azCmbdYJxZ97N+NIGYLm7+89wjDg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-12T08:54:26.817240Z","bundle_sha256":"6c549ceef0113c70ce27d77b13cfede2df926152777f081e10c728f0d52d8102"}}