{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:TPLINE6IZX4T5BFPA7ETJUPKPJ","short_pith_number":"pith:TPLINE6I","canonical_record":{"source":{"id":"2403.14459","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-03-21T15:06:14Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"20c4f93362c44c824034d938a4cf2f00730b7b571817503bb92f04f377d72a35","abstract_canon_sha256":"061568474c70d177fb859bf21af52385da0c79bb176edc7382cd0ab2b7367879"},"schema_version":"1.0"},"canonical_sha256":"9bd68693c8cdf93e84af07c934d1ea7a4435b94c05342138c12b232e48ed15a7","source":{"kind":"arxiv","id":"2403.14459","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2403.14459","created_at":"2026-07-05T11:41:33Z"},{"alias_kind":"arxiv_version","alias_value":"2403.14459v2","created_at":"2026-07-05T11:41:33Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.14459","created_at":"2026-07-05T11:41:33Z"},{"alias_kind":"pith_short_12","alias_value":"TPLINE6IZX4T","created_at":"2026-07-05T11:41:33Z"},{"alias_kind":"pith_short_16","alias_value":"TPLINE6IZX4T5BFP","created_at":"2026-07-05T11:41:33Z"},{"alias_kind":"pith_short_8","alias_value":"TPLINE6I","created_at":"2026-07-05T11:41:33Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:TPLINE6IZX4T5BFPA7ETJUPKPJ","target":"record","payload":{"canonical_record":{"source":{"id":"2403.14459","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-03-21T15:06:14Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"20c4f93362c44c824034d938a4cf2f00730b7b571817503bb92f04f377d72a35","abstract_canon_sha256":"061568474c70d177fb859bf21af52385da0c79bb176edc7382cd0ab2b7367879"},"schema_version":"1.0"},"canonical_sha256":"9bd68693c8cdf93e84af07c934d1ea7a4435b94c05342138c12b232e48ed15a7","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:41:33.791587Z","signature_b64":"SYtDTP6Z4gSe165IjrTJba7Gsa1SGWgJZXTmDeEjq/IQbHEHumm58tJUM8BNv8lIAvPGFGplcEOzKRGBtN5pAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9bd68693c8cdf93e84af07c934d1ea7a4435b94c05342138c12b232e48ed15a7","last_reissued_at":"2026-07-05T11:41:33.790996Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:41:33.790996Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2403.14459","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":"2026-07-05T11:41:33Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"6kv6mhRLS/Glhf6wNiPn9h+UkKYTLJf2lvFBkxfaV7B7WGDtYcHPstz72IP0fX4+y06v5sEUUbTSc3GUF1a6BA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-10T11:49:04.673884Z"},"content_sha256":"78dbad0ab137f2971ae24a1a9708c847ae19e23859d01efc5f0b4ade778c03de","schema_version":"1.0","event_id":"sha256:78dbad0ab137f2971ae24a1a9708c847ae19e23859d01efc5f0b4ade778c03de"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:TPLINE6IZX4T5BFPA7ETJUPKPJ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Multi-Level Explanations for Generative Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Amit Dhurandhar, Dennis Wei, Hendrik Strobelt, Hyo Jin Do, Karthikeyan Natesan Ramamurthy, Lucas Monteiro Paes, Manish Nagireddy, Prasanna Sattigeri, Ronny Luss, Soumya Ghosh, Werner Geyer","submitted_at":"2024-03-21T15:06:14Z","abstract_excerpt":"Despite the increasing use of large language models (LLMs) for context-grounded tasks like summarization and question-answering, understanding what makes an LLM produce a certain response is challenging. We propose Multi-Level Explanations for Generative Language Models (MExGen), a technique to provide explanations for context-grounded text generation. MExGen assigns scores to parts of the context to quantify their influence on the model's output. It extends attribution methods like LIME and SHAP to LLMs used in context-grounded tasks where (1) inference cost is high, (2) input text is long, a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.14459","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/2403.14459/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":"2026-07-05T11:41:33Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"mY0ksW37yaZDlHoyQmdRNG2essnpJGZ5xyiwFUwd8o30hwp2lAQ5qfkP/P4XZUK2+m9OHCR0QlDRxAyHT2r0Aw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-10T11:49:04.674393Z"},"content_sha256":"e297e90331649507a98241eaaa738281d7f912dd501b176f8406341ea8d7e661","schema_version":"1.0","event_id":"sha256:e297e90331649507a98241eaaa738281d7f912dd501b176f8406341ea8d7e661"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/TPLINE6IZX4T5BFPA7ETJUPKPJ/bundle.json","state_url":"https://pith.science/pith/TPLINE6IZX4T5BFPA7ETJUPKPJ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/TPLINE6IZX4T5BFPA7ETJUPKPJ/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":"2026-08-10T11:49:04Z","links":{"resolver":"https://pith.science/pith/TPLINE6IZX4T5BFPA7ETJUPKPJ","bundle":"https://pith.science/pith/TPLINE6IZX4T5BFPA7ETJUPKPJ/bundle.json","state":"https://pith.science/pith/TPLINE6IZX4T5BFPA7ETJUPKPJ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/TPLINE6IZX4T5BFPA7ETJUPKPJ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:TPLINE6IZX4T5BFPA7ETJUPKPJ","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":"061568474c70d177fb859bf21af52385da0c79bb176edc7382cd0ab2b7367879","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-03-21T15:06:14Z","title_canon_sha256":"20c4f93362c44c824034d938a4cf2f00730b7b571817503bb92f04f377d72a35"},"schema_version":"1.0","source":{"id":"2403.14459","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2403.14459","created_at":"2026-07-05T11:41:33Z"},{"alias_kind":"arxiv_version","alias_value":"2403.14459v2","created_at":"2026-07-05T11:41:33Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.14459","created_at":"2026-07-05T11:41:33Z"},{"alias_kind":"pith_short_12","alias_value":"TPLINE6IZX4T","created_at":"2026-07-05T11:41:33Z"},{"alias_kind":"pith_short_16","alias_value":"TPLINE6IZX4T5BFP","created_at":"2026-07-05T11:41:33Z"},{"alias_kind":"pith_short_8","alias_value":"TPLINE6I","created_at":"2026-07-05T11:41:33Z"}],"graph_snapshots":[{"event_id":"sha256:e297e90331649507a98241eaaa738281d7f912dd501b176f8406341ea8d7e661","target":"graph","created_at":"2026-07-05T11:41:33Z","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/2403.14459/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Despite the increasing use of large language models (LLMs) for context-grounded tasks like summarization and question-answering, understanding what makes an LLM produce a certain response is challenging. We propose Multi-Level Explanations for Generative Language Models (MExGen), a technique to provide explanations for context-grounded text generation. MExGen assigns scores to parts of the context to quantify their influence on the model's output. It extends attribution methods like LIME and SHAP to LLMs used in context-grounded tasks where (1) inference cost is high, (2) input text is long, a","authors_text":"Amit Dhurandhar, Dennis Wei, Hendrik Strobelt, Hyo Jin Do, Karthikeyan Natesan Ramamurthy, Lucas Monteiro Paes, Manish Nagireddy, Prasanna Sattigeri, Ronny Luss, Soumya Ghosh, Werner Geyer","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-03-21T15:06:14Z","title":"Multi-Level Explanations for Generative Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.14459","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":"sha256:78dbad0ab137f2971ae24a1a9708c847ae19e23859d01efc5f0b4ade778c03de","target":"record","created_at":"2026-07-05T11:41:33Z","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":"061568474c70d177fb859bf21af52385da0c79bb176edc7382cd0ab2b7367879","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-03-21T15:06:14Z","title_canon_sha256":"20c4f93362c44c824034d938a4cf2f00730b7b571817503bb92f04f377d72a35"},"schema_version":"1.0","source":{"id":"2403.14459","kind":"arxiv","version":2}},"canonical_sha256":"9bd68693c8cdf93e84af07c934d1ea7a4435b94c05342138c12b232e48ed15a7","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"9bd68693c8cdf93e84af07c934d1ea7a4435b94c05342138c12b232e48ed15a7","first_computed_at":"2026-07-05T11:41:33.790996Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:41:33.790996Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"SYtDTP6Z4gSe165IjrTJba7Gsa1SGWgJZXTmDeEjq/IQbHEHumm58tJUM8BNv8lIAvPGFGplcEOzKRGBtN5pAw==","signature_status":"signed_v1","signed_at":"2026-07-05T11:41:33.791587Z","signed_message":"canonical_sha256_bytes"},"source_id":"2403.14459","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:78dbad0ab137f2971ae24a1a9708c847ae19e23859d01efc5f0b4ade778c03de","sha256:e297e90331649507a98241eaaa738281d7f912dd501b176f8406341ea8d7e661"],"state_sha256":"659ae57317fc0b590704ca0d1e6988fb9f5fe98218010f5b969be29fa6d28f15"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"sf2jKLCK+NpN+HZt274nAnPsv1DnnPcrKPNI9LI+AvSNDw0DPcVjf0UvI/fOeDPHTsqCP4r9KzCrJ3ETE1cJDg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-10T11:49:04.679790Z","bundle_sha256":"214caacdd23c11b4a9cbc036ce49e93ce44377567afb31fb70d2522234ad77ab"}}