{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:2UVJMP4O2M3DUYVRISAWJJHSUE","short_pith_number":"pith:2UVJMP4O","canonical_record":{"source":{"id":"2503.16870","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-21T05:58:18Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"13ea0ad45d71e1cfbae70c3f6403a32dd6d4b0d341b214d72b6adc424d1e7981","abstract_canon_sha256":"70b889e21366465fea4eb846d654c26ab1d9bdfa07734ba457d8be5ded606249"},"schema_version":"1.0"},"canonical_sha256":"d52a963f8ed3363a62b1448164a4f2a10b0f012c904161cf987bf3c16190e01f","source":{"kind":"arxiv","id":"2503.16870","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2503.16870","created_at":"2026-07-05T11:42:27Z"},{"alias_kind":"arxiv_version","alias_value":"2503.16870v2","created_at":"2026-07-05T11:42:27Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.16870","created_at":"2026-07-05T11:42:27Z"},{"alias_kind":"pith_short_12","alias_value":"2UVJMP4O2M3D","created_at":"2026-07-05T11:42:27Z"},{"alias_kind":"pith_short_16","alias_value":"2UVJMP4O2M3DUYVR","created_at":"2026-07-05T11:42:27Z"},{"alias_kind":"pith_short_8","alias_value":"2UVJMP4O","created_at":"2026-07-05T11:42:27Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:2UVJMP4O2M3DUYVRISAWJJHSUE","target":"record","payload":{"canonical_record":{"source":{"id":"2503.16870","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-21T05:58:18Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"13ea0ad45d71e1cfbae70c3f6403a32dd6d4b0d341b214d72b6adc424d1e7981","abstract_canon_sha256":"70b889e21366465fea4eb846d654c26ab1d9bdfa07734ba457d8be5ded606249"},"schema_version":"1.0"},"canonical_sha256":"d52a963f8ed3363a62b1448164a4f2a10b0f012c904161cf987bf3c16190e01f","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:42:27.673872Z","signature_b64":"n2hIQwlDLiCf6wRESkHFYSKReOvlMZkkdOYxvxhhehqT7eYTCWZpYSrpt0Ayx9LUFBhjcaRZSjOSSey1l8uKBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d52a963f8ed3363a62b1448164a4f2a10b0f012c904161cf987bf3c16190e01f","last_reissued_at":"2026-07-05T11:42:27.673323Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:42:27.673323Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2503.16870","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":"2026-07-05T11:42:27Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"CbNWScd9XpHa/RbFzn9ALtJU27zWAYxPFw2lk7A1kUDjf31RO4UpSIkb06F1/b2WHl7Q50vrtTDcqVpw4H8/Dw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-01T22:29:33.165580Z"},"content_sha256":"1c8b7d67ccf2062c627b8ebe30e55ea64d2c24091df00b43b67539d494c98a23","schema_version":"1.0","event_id":"sha256:1c8b7d67ccf2062c627b8ebe30e55ea64d2c24091df00b43b67539d494c98a23"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:2UVJMP4O2M3DUYVRISAWJJHSUE","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Sparse Logit Sampling: Accelerating Knowledge Distillation in LLMs","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Akhil Kedia, Anshumann, Haejun Lee, Jinwoo Ahn, Joohyung Lee, Kangwook Lee, Mohd Abbas Zaidi, Taehwak Kwon","submitted_at":"2025-03-21T05:58:18Z","abstract_excerpt":"Knowledge distillation can be a cost-effective technique to distill knowledge in Large Language Models, if the teacher output logits can be pre-computed and cached. However, successfully applying this to pre-training remains largely unexplored. In this work, we prove that naive approaches for sparse knowledge distillation such as caching Top-K probabilities, while intuitive, provide biased estimates of teacher probability distribution to the student, resulting in suboptimal performance and calibration. We propose an importance-sampling-based method `Random Sampling Knowledge Distillation', whi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.16870","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/2503.16870/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":"2026-07-05T11:42:27Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"o3eF7+ENk6Eloi+3fT2Q/kPiYhqyTo9JnbNbXzhlTcQuGP6rXpUX8K3Wfj9fxQFleEEVQV2ShVWjjPF+su1yCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-01T22:29:33.166089Z"},"content_sha256":"b8efed69b718a11cfb969f6aebe472c3fb725c7f976479ba758ca8be00b5b4df","schema_version":"1.0","event_id":"sha256:b8efed69b718a11cfb969f6aebe472c3fb725c7f976479ba758ca8be00b5b4df"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/2UVJMP4O2M3DUYVRISAWJJHSUE/bundle.json","state_url":"https://pith.science/pith/2UVJMP4O2M3DUYVRISAWJJHSUE/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/2UVJMP4O2M3DUYVRISAWJJHSUE/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":"2026-08-01T22:29:33Z","links":{"resolver":"https://pith.science/pith/2UVJMP4O2M3DUYVRISAWJJHSUE","bundle":"https://pith.science/pith/2UVJMP4O2M3DUYVRISAWJJHSUE/bundle.json","state":"https://pith.science/pith/2UVJMP4O2M3DUYVRISAWJJHSUE/state.json","well_known_bundle":"https://pith.science/.well-known/pith/2UVJMP4O2M3DUYVRISAWJJHSUE/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:2UVJMP4O2M3DUYVRISAWJJHSUE","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":"70b889e21366465fea4eb846d654c26ab1d9bdfa07734ba457d8be5ded606249","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-21T05:58:18Z","title_canon_sha256":"13ea0ad45d71e1cfbae70c3f6403a32dd6d4b0d341b214d72b6adc424d1e7981"},"schema_version":"1.0","source":{"id":"2503.16870","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2503.16870","created_at":"2026-07-05T11:42:27Z"},{"alias_kind":"arxiv_version","alias_value":"2503.16870v2","created_at":"2026-07-05T11:42:27Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.16870","created_at":"2026-07-05T11:42:27Z"},{"alias_kind":"pith_short_12","alias_value":"2UVJMP4O2M3D","created_at":"2026-07-05T11:42:27Z"},{"alias_kind":"pith_short_16","alias_value":"2UVJMP4O2M3DUYVR","created_at":"2026-07-05T11:42:27Z"},{"alias_kind":"pith_short_8","alias_value":"2UVJMP4O","created_at":"2026-07-05T11:42:27Z"}],"graph_snapshots":[{"event_id":"sha256:b8efed69b718a11cfb969f6aebe472c3fb725c7f976479ba758ca8be00b5b4df","target":"graph","created_at":"2026-07-05T11:42:27Z","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/2503.16870/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Knowledge distillation can be a cost-effective technique to distill knowledge in Large Language Models, if the teacher output logits can be pre-computed and cached. However, successfully applying this to pre-training remains largely unexplored. In this work, we prove that naive approaches for sparse knowledge distillation such as caching Top-K probabilities, while intuitive, provide biased estimates of teacher probability distribution to the student, resulting in suboptimal performance and calibration. We propose an importance-sampling-based method `Random Sampling Knowledge Distillation', whi","authors_text":"Akhil Kedia, Anshumann, Haejun Lee, Jinwoo Ahn, Joohyung Lee, Kangwook Lee, Mohd Abbas Zaidi, Taehwak Kwon","cross_cats":["cs.AI","cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-21T05:58:18Z","title":"Sparse Logit Sampling: Accelerating Knowledge Distillation in LLMs"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.16870","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":"sha256:1c8b7d67ccf2062c627b8ebe30e55ea64d2c24091df00b43b67539d494c98a23","target":"record","created_at":"2026-07-05T11:42:27Z","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":"70b889e21366465fea4eb846d654c26ab1d9bdfa07734ba457d8be5ded606249","cross_cats_sorted":["cs.AI","cs.CL"],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-21T05:58:18Z","title_canon_sha256":"13ea0ad45d71e1cfbae70c3f6403a32dd6d4b0d341b214d72b6adc424d1e7981"},"schema_version":"1.0","source":{"id":"2503.16870","kind":"arxiv","version":2}},"canonical_sha256":"d52a963f8ed3363a62b1448164a4f2a10b0f012c904161cf987bf3c16190e01f","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"d52a963f8ed3363a62b1448164a4f2a10b0f012c904161cf987bf3c16190e01f","first_computed_at":"2026-07-05T11:42:27.673323Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:42:27.673323Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"n2hIQwlDLiCf6wRESkHFYSKReOvlMZkkdOYxvxhhehqT7eYTCWZpYSrpt0Ayx9LUFBhjcaRZSjOSSey1l8uKBQ==","signature_status":"signed_v1","signed_at":"2026-07-05T11:42:27.673872Z","signed_message":"canonical_sha256_bytes"},"source_id":"2503.16870","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:1c8b7d67ccf2062c627b8ebe30e55ea64d2c24091df00b43b67539d494c98a23","sha256:b8efed69b718a11cfb969f6aebe472c3fb725c7f976479ba758ca8be00b5b4df"],"state_sha256":"f019ae00eb2419670fabe3a83c9d5eadedebc412358089bdbaf0d21c584053b5"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"YY6MSr0JZ9wYqOLT/hHMKmoyuzKjShR9npDA206Q03e45YIChwAAb4Wy9Wuie2cRoEgABWSatVfZqc831gf0Dw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-01T22:29:33.169496Z","bundle_sha256":"9ed174571ee8179a965651d72a7d87bd297579ffc59399cadf96caf3156cd24e"}}