{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:GFTYHFRX5C6MGUBUO2VYTJXWIY","short_pith_number":"pith:GFTYHFRX","schema_version":"1.0","canonical_sha256":"3167839637e8bcc3503476ab89a6f6462098ea487e700364a1f19548b651f756","source":{"kind":"arxiv","id":"2404.19319","version":1},"attestation_state":"computed","paper":{"title":"Knowledge Distillation vs. Pretraining from Scratch under a Fixed (Computation) Budget","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Fabian David Schmidt, Goran Glava\\v{s}, Katharina von der Wense, Minh Duc Bui","submitted_at":"2024-04-30T07:40:35Z","abstract_excerpt":"Compared to standard language model (LM) pretraining (i.e., from scratch), Knowledge Distillation (KD) entails an additional forward pass through a teacher model that is typically substantially larger than the target student model. As such, KD in LM pretraining materially slows down throughput of pretraining instances vis-a-vis pretraining from scratch. Scaling laws of LM pretraining suggest that smaller models can close the gap to larger counterparts if trained on more data (i.e., processing more tokens)-and under a fixed computation budget, smaller models are able be process more data than l"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2404.19319","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-04-30T07:40:35Z","cross_cats_sorted":[],"title_canon_sha256":"733bb55dc92505d51b849600603fc12735e145410d8eda62286a416a32453d85","abstract_canon_sha256":"0958ef01fcc39cb31856147e1517d27482f046f87f71764ca7b8712054144e2b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:13:41.464871Z","signature_b64":"CFwmolWcodqiFUxgz6LGHWIfZa8uzHdNtl0DoyL60A0jdsTmc6lZmetXh1lTcwxq3W4/mAEjfi3/ZetCW7qRAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3167839637e8bcc3503476ab89a6f6462098ea487e700364a1f19548b651f756","last_reissued_at":"2026-07-05T08:13:41.464412Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:13:41.464412Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Knowledge Distillation vs. Pretraining from Scratch under a Fixed (Computation) Budget","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Fabian David Schmidt, Goran Glava\\v{s}, Katharina von der Wense, Minh Duc Bui","submitted_at":"2024-04-30T07:40:35Z","abstract_excerpt":"Compared to standard language model (LM) pretraining (i.e., from scratch), Knowledge Distillation (KD) entails an additional forward pass through a teacher model that is typically substantially larger than the target student model. As such, KD in LM pretraining materially slows down throughput of pretraining instances vis-a-vis pretraining from scratch. Scaling laws of LM pretraining suggest that smaller models can close the gap to larger counterparts if trained on more data (i.e., processing more tokens)-and under a fixed computation budget, smaller models are able be process more data than l"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.19319","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/2404.19319/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2404.19319","created_at":"2026-07-05T08:13:41.464469+00:00"},{"alias_kind":"arxiv_version","alias_value":"2404.19319v1","created_at":"2026-07-05T08:13:41.464469+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.19319","created_at":"2026-07-05T08:13:41.464469+00:00"},{"alias_kind":"pith_short_12","alias_value":"GFTYHFRX5C6M","created_at":"2026-07-05T08:13:41.464469+00:00"},{"alias_kind":"pith_short_16","alias_value":"GFTYHFRX5C6MGUBU","created_at":"2026-07-05T08:13:41.464469+00:00"},{"alias_kind":"pith_short_8","alias_value":"GFTYHFRX","created_at":"2026-07-05T08:13:41.464469+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2506.11040","citing_title":"Large Language models for Time Series Analysis: Techniques, Applications, and Challenges","ref_index":72,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GFTYHFRX5C6MGUBUO2VYTJXWIY","json":"https://pith.science/pith/GFTYHFRX5C6MGUBUO2VYTJXWIY.json","graph_json":"https://pith.science/api/pith-number/GFTYHFRX5C6MGUBUO2VYTJXWIY/graph.json","events_json":"https://pith.science/api/pith-number/GFTYHFRX5C6MGUBUO2VYTJXWIY/events.json","paper":"https://pith.science/paper/GFTYHFRX"},"agent_actions":{"view_html":"https://pith.science/pith/GFTYHFRX5C6MGUBUO2VYTJXWIY","download_json":"https://pith.science/pith/GFTYHFRX5C6MGUBUO2VYTJXWIY.json","view_paper":"https://pith.science/paper/GFTYHFRX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2404.19319&json=true","fetch_graph":"https://pith.science/api/pith-number/GFTYHFRX5C6MGUBUO2VYTJXWIY/graph.json","fetch_events":"https://pith.science/api/pith-number/GFTYHFRX5C6MGUBUO2VYTJXWIY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GFTYHFRX5C6MGUBUO2VYTJXWIY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GFTYHFRX5C6MGUBUO2VYTJXWIY/action/storage_attestation","attest_author":"https://pith.science/pith/GFTYHFRX5C6MGUBUO2VYTJXWIY/action/author_attestation","sign_citation":"https://pith.science/pith/GFTYHFRX5C6MGUBUO2VYTJXWIY/action/citation_signature","submit_replication":"https://pith.science/pith/GFTYHFRX5C6MGUBUO2VYTJXWIY/action/replication_record"}},"created_at":"2026-07-05T08:13:41.464469+00:00","updated_at":"2026-07-05T08:13:41.464469+00:00"}