{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:RWZNSQPCB5RPNHWPA5STL4FYYK","short_pith_number":"pith:RWZNSQPC","schema_version":"1.0","canonical_sha256":"8db2d941e20f62f69ecf076535f0b8c28024d1d5f578ef165a29250c23acf9db","source":{"kind":"arxiv","id":"1908.08962","version":2},"attestation_state":"computed","paper":{"title":"Well-Read Students Learn Better: On the Importance of Pre-training Compact Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Iulia Turc, Kenton Lee, Kristina Toutanova, Ming-Wei Chang","submitted_at":"2019-08-23T18:02:05Z","abstract_excerpt":"Recent developments in natural language representations have been accompanied by large and expensive models that leverage vast amounts of general-domain text through self-supervised pre-training. Due to the cost of applying such models to down-stream tasks, several model compression techniques on pre-trained language representations have been proposed (Sun et al., 2019; Sanh, 2019). However, surprisingly, the simple baseline of just pre-training and fine-tuning compact models has been overlooked. In this paper, we first show that pre-training remains important in the context of smaller archite"},"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":"1908.08962","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2019-08-23T18:02:05Z","cross_cats_sorted":[],"title_canon_sha256":"ed18a0abb24006a8c85c9bf78be0c9636abf422ca8d8ee1f059419278993ef83","abstract_canon_sha256":"1ae26a86abd8691c953b599d6106ede2c6d8c6ab2530f4db0d915c9e8397f372"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:07:24.565408Z","signature_b64":"pLDjTnDtZdbR0mWL7g4WLgtUHbpSQexvGP/VUKI/dZ78cxefQ2FHB1w0Uh1LD9qRmdix/q8N9gQbEcWJrCLXCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8db2d941e20f62f69ecf076535f0b8c28024d1d5f578ef165a29250c23acf9db","last_reissued_at":"2026-07-05T00:07:24.564920Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:07:24.564920Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Well-Read Students Learn Better: On the Importance of Pre-training Compact Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Iulia Turc, Kenton Lee, Kristina Toutanova, Ming-Wei Chang","submitted_at":"2019-08-23T18:02:05Z","abstract_excerpt":"Recent developments in natural language representations have been accompanied by large and expensive models that leverage vast amounts of general-domain text through self-supervised pre-training. Due to the cost of applying such models to down-stream tasks, several model compression techniques on pre-trained language representations have been proposed (Sun et al., 2019; Sanh, 2019). However, surprisingly, the simple baseline of just pre-training and fine-tuning compact models has been overlooked. In this paper, we first show that pre-training remains important in the context of smaller archite"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1908.08962","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/1908.08962/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":"1908.08962","created_at":"2026-07-05T00:07:24.564980+00:00"},{"alias_kind":"arxiv_version","alias_value":"1908.08962v2","created_at":"2026-07-05T00:07:24.564980+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1908.08962","created_at":"2026-07-05T00:07:24.564980+00:00"},{"alias_kind":"pith_short_12","alias_value":"RWZNSQPCB5RP","created_at":"2026-07-05T00:07:24.564980+00:00"},{"alias_kind":"pith_short_16","alias_value":"RWZNSQPCB5RPNHWP","created_at":"2026-07-05T00:07:24.564980+00:00"},{"alias_kind":"pith_short_8","alias_value":"RWZNSQPC","created_at":"2026-07-05T00:07:24.564980+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":12,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.22327","citing_title":"Geometry-Aware Online Scheduling for LLM Serving: From Theoretical Bound to System Practice","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2606.21240","citing_title":"DIPBox: A Multi-scale Testing Framework for Tracking Dataset Regeneration","ref_index":68,"is_internal_anchor":false},{"citing_arxiv_id":"2606.21593","citing_title":"Geometric and Information Compression of Representations in Deep Learning","ref_index":49,"is_internal_anchor":false},{"citing_arxiv_id":"2606.02100","citing_title":"PortBERT: Navigating the Depths of Portuguese Language Models","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2502.05075","citing_title":"Discrepancies are Virtue: Weak-to-Strong Generalization through Lens of Intrinsic Dimension","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2210.08933","citing_title":"DiffuSeq: Sequence to Sequence Text Generation with Diffusion Models","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2101.02388","citing_title":"Knowledge Distillation in Iterative Generative Models for Improved Sampling Speed","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2603.20910","citing_title":"LLM-ODE: Data-driven Discovery of Dynamical Systems with Large Language Models","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"1909.11942","citing_title":"ALBERT: A Lite BERT for Self-supervised Learning of Language Representations","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11695","citing_title":"Emergent Communication between Heterogeneous Visual Agents through Decentralized Learning","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"1910.01108","citing_title":"DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07297","citing_title":"Spectrum-Adaptive Generalization Bounds for Trained Deep Transformers","ref_index":27,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RWZNSQPCB5RPNHWPA5STL4FYYK","json":"https://pith.science/pith/RWZNSQPCB5RPNHWPA5STL4FYYK.json","graph_json":"https://pith.science/api/pith-number/RWZNSQPCB5RPNHWPA5STL4FYYK/graph.json","events_json":"https://pith.science/api/pith-number/RWZNSQPCB5RPNHWPA5STL4FYYK/events.json","paper":"https://pith.science/paper/RWZNSQPC"},"agent_actions":{"view_html":"https://pith.science/pith/RWZNSQPCB5RPNHWPA5STL4FYYK","download_json":"https://pith.science/pith/RWZNSQPCB5RPNHWPA5STL4FYYK.json","view_paper":"https://pith.science/paper/RWZNSQPC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1908.08962&json=true","fetch_graph":"https://pith.science/api/pith-number/RWZNSQPCB5RPNHWPA5STL4FYYK/graph.json","fetch_events":"https://pith.science/api/pith-number/RWZNSQPCB5RPNHWPA5STL4FYYK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RWZNSQPCB5RPNHWPA5STL4FYYK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RWZNSQPCB5RPNHWPA5STL4FYYK/action/storage_attestation","attest_author":"https://pith.science/pith/RWZNSQPCB5RPNHWPA5STL4FYYK/action/author_attestation","sign_citation":"https://pith.science/pith/RWZNSQPCB5RPNHWPA5STL4FYYK/action/citation_signature","submit_replication":"https://pith.science/pith/RWZNSQPCB5RPNHWPA5STL4FYYK/action/replication_record"}},"created_at":"2026-07-05T00:07:24.564980+00:00","updated_at":"2026-07-05T00:07:24.564980+00:00"}