{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:ZH6LT6B2HQ2DOT7C6HT6UZASM4","short_pith_number":"pith:ZH6LT6B2","schema_version":"1.0","canonical_sha256":"c9fcb9f83a3c34374fe2f1e7ea6412673c011ca8e99f5dac7b34035a9dcee493","source":{"kind":"arxiv","id":"2504.15027","version":1},"attestation_state":"computed","paper":{"title":"DistilQwen2.5: Industrial Practices of Training Distilled Open Lightweight Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chengyu Wang, Junbing Yan, Jun Huang, Yuanhao Yue","submitted_at":"2025-04-21T11:26:02Z","abstract_excerpt":"Enhancing computational efficiency and reducing deployment costs for large language models (LLMs) have become critical challenges in various resource-constrained scenarios. In this work, we present DistilQwen2.5, a family of distilled, lightweight LLMs derived from the public Qwen2.5 models. These distilled models exhibit enhanced instruction-following capabilities compared to the original models based on a series of distillation techniques that incorporate knowledge from much larger LLMs. In our industrial practice, we first leverage powerful proprietary LLMs with varying capacities as multi-"},"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":"2504.15027","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-04-21T11:26:02Z","cross_cats_sorted":[],"title_canon_sha256":"22ca69a23e847b0793e27affe29b3f3d1cd63c29900d8479fcbd10a8e43a4894","abstract_canon_sha256":"d7b217f6f6356c564c79adfefa2675d23bbc46442e1f669a94245ec35d1e7839"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:51:55.165075Z","signature_b64":"Awrs3NalcTlEUdXpLFD1UIOMXdrdtnLWHsN8xJGT0ktD74bcOpOF2ATcBcu4mIik0uifI2RbGaW2q+lmUdaGDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c9fcb9f83a3c34374fe2f1e7ea6412673c011ca8e99f5dac7b34035a9dcee493","last_reissued_at":"2026-07-05T10:51:55.164593Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:51:55.164593Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DistilQwen2.5: Industrial Practices of Training Distilled Open Lightweight Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chengyu Wang, Junbing Yan, Jun Huang, Yuanhao Yue","submitted_at":"2025-04-21T11:26:02Z","abstract_excerpt":"Enhancing computational efficiency and reducing deployment costs for large language models (LLMs) have become critical challenges in various resource-constrained scenarios. In this work, we present DistilQwen2.5, a family of distilled, lightweight LLMs derived from the public Qwen2.5 models. These distilled models exhibit enhanced instruction-following capabilities compared to the original models based on a series of distillation techniques that incorporate knowledge from much larger LLMs. In our industrial practice, we first leverage powerful proprietary LLMs with varying capacities as multi-"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.15027","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/2504.15027/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":"2504.15027","created_at":"2026-07-05T10:51:55.164654+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.15027v1","created_at":"2026-07-05T10:51:55.164654+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.15027","created_at":"2026-07-05T10:51:55.164654+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZH6LT6B2HQ2D","created_at":"2026-07-05T10:51:55.164654+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZH6LT6B2HQ2DOT7C","created_at":"2026-07-05T10:51:55.164654+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZH6LT6B2","created_at":"2026-07-05T10:51:55.164654+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.11629","citing_title":"OmniThoughtVis: A Scalable Distillation Pipeline for Deployable Multimodal Reasoning Models","ref_index":22,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZH6LT6B2HQ2DOT7C6HT6UZASM4","json":"https://pith.science/pith/ZH6LT6B2HQ2DOT7C6HT6UZASM4.json","graph_json":"https://pith.science/api/pith-number/ZH6LT6B2HQ2DOT7C6HT6UZASM4/graph.json","events_json":"https://pith.science/api/pith-number/ZH6LT6B2HQ2DOT7C6HT6UZASM4/events.json","paper":"https://pith.science/paper/ZH6LT6B2"},"agent_actions":{"view_html":"https://pith.science/pith/ZH6LT6B2HQ2DOT7C6HT6UZASM4","download_json":"https://pith.science/pith/ZH6LT6B2HQ2DOT7C6HT6UZASM4.json","view_paper":"https://pith.science/paper/ZH6LT6B2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.15027&json=true","fetch_graph":"https://pith.science/api/pith-number/ZH6LT6B2HQ2DOT7C6HT6UZASM4/graph.json","fetch_events":"https://pith.science/api/pith-number/ZH6LT6B2HQ2DOT7C6HT6UZASM4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZH6LT6B2HQ2DOT7C6HT6UZASM4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZH6LT6B2HQ2DOT7C6HT6UZASM4/action/storage_attestation","attest_author":"https://pith.science/pith/ZH6LT6B2HQ2DOT7C6HT6UZASM4/action/author_attestation","sign_citation":"https://pith.science/pith/ZH6LT6B2HQ2DOT7C6HT6UZASM4/action/citation_signature","submit_replication":"https://pith.science/pith/ZH6LT6B2HQ2DOT7C6HT6UZASM4/action/replication_record"}},"created_at":"2026-07-05T10:51:55.164654+00:00","updated_at":"2026-07-05T10:51:55.164654+00:00"}