{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:KPASI7VIC6IYHQLPWCAXGHSEUC","short_pith_number":"pith:KPASI7VI","schema_version":"1.0","canonical_sha256":"53c1247ea8179183c16fb081731e44a0a19c9adc8a47da9dccf78fc37a6aee80","source":{"kind":"arxiv","id":"2404.07965","version":4},"attestation_state":"computed","paper":{"title":"Rho-1: Not All Tokens Are What You Need","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Chen Lin, Jian Jiao, Nan Duan, Ruochen Xu, Weizhu Chen, Xiao Liu, Yelong Shen, Yeyun Gong, Yujiu Yang, Zhenghao Lin, Zhibin Gou","submitted_at":"2024-04-11T17:52:01Z","abstract_excerpt":"Previous language model pre-training methods have uniformly applied a next-token prediction loss to all training tokens. Challenging this norm, we posit that \"9l training\". Our initial analysis examines token-level training dynamics of language model, revealing distinct loss patterns for different tokens. Leveraging these insights, we introduce a new language model called Rho-1. Unlike traditional LMs that learn to predict every next token in a corpus, Rho-1 employs Selective Language Modeling (SLM), which selectively trains on useful tokens that aligned with the desired distribution. This app"},"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":"2404.07965","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-04-11T17:52:01Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"15f9dc3caa54332cbfc969384b9d38a01b43a7cd0328e61c83faf603131880aa","abstract_canon_sha256":"48a4f594aa828e140f6546f91ed888b292f938a96a61be2a787ceec2d2046676"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:58:19.860801Z","signature_b64":"CZOpJVMTUwoqTqeqMMVLA05iws5WCrqFjN2UvouKWfyUGG2DjnvlQMrykW3y7Nw2Opn6y7GLQ5M89TxsnQv9DQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"53c1247ea8179183c16fb081731e44a0a19c9adc8a47da9dccf78fc37a6aee80","last_reissued_at":"2026-07-05T09:58:19.860228Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:58:19.860228Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Rho-1: Not All Tokens Are What You Need","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Chen Lin, Jian Jiao, Nan Duan, Ruochen Xu, Weizhu Chen, Xiao Liu, Yelong Shen, Yeyun Gong, Yujiu Yang, Zhenghao Lin, Zhibin Gou","submitted_at":"2024-04-11T17:52:01Z","abstract_excerpt":"Previous language model pre-training methods have uniformly applied a next-token prediction loss to all training tokens. Challenging this norm, we posit that \"9l training\". Our initial analysis examines token-level training dynamics of language model, revealing distinct loss patterns for different tokens. Leveraging these insights, we introduce a new language model called Rho-1. Unlike traditional LMs that learn to predict every next token in a corpus, Rho-1 employs Selective Language Modeling (SLM), which selectively trains on useful tokens that aligned with the desired distribution. This app"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.07965","kind":"arxiv","version":4},"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/2404.07965/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":"2404.07965","created_at":"2026-07-05T09:58:19.860308+00:00"},{"alias_kind":"arxiv_version","alias_value":"2404.07965v4","created_at":"2026-07-05T09:58:19.860308+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.07965","created_at":"2026-07-05T09:58:19.860308+00:00"},{"alias_kind":"pith_short_12","alias_value":"KPASI7VIC6IY","created_at":"2026-07-05T09:58:19.860308+00:00"},{"alias_kind":"pith_short_16","alias_value":"KPASI7VIC6IYHQLP","created_at":"2026-07-05T09:58:19.860308+00:00"},{"alias_kind":"pith_short_8","alias_value":"KPASI7VI","created_at":"2026-07-05T09:58:19.860308+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":15,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.18286","citing_title":"CODEBLOCK: Learning to Supervise Code at the Right Granularity","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2502.12272","citing_title":"Learning to Reason at the Frontier of Learnability","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2510.07962","citing_title":"LightReasoner: Can Small Language Models Teach Large Language Models Reasoning?","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2601.10348","citing_title":"Training-Trajectory-Aware Token Selection","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2507.21420","citing_title":"ReGATE: Learning Faster and Better with Fewer Tokens in MLLMs","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2406.18629","citing_title":"Step-DPO: Step-wise Preference Optimization for Long-chain Reasoning of LLMs","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2508.15229","citing_title":"VocabTailor: Dynamic Vocabulary Selection for Downstream Tasks in Small Language Models","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2509.12089","citing_title":"RadarPLM: Adapting Pre-trained Language Models for Marine Radar Target Detection by Selective Fine-tuning","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2509.20863","citing_title":"GIFT: Guided Importance-Aware Fine-Tuning for Diffusion Language Models","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2509.20863","citing_title":"GIFT: Guided Importance-Aware Fine-Tuning for Diffusion Language Models","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2406.11794","citing_title":"DataComp-LM: In search of the next generation of training sets for language models","ref_index":108,"is_internal_anchor":false},{"citing_arxiv_id":"2602.20816","citing_title":"Don't Ignore the Tail: Decoupling top-K Probabilities for Efficient Language Model Distillation","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2603.01097","citing_title":"Understanding LoRA as Knowledge Memory: An Empirical Analysis","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2604.22374","citing_title":"Selective Contrastive Learning For Gloss Free Sign Language Translation","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17884","citing_title":"SPREG: Structured Plan Repair with Entropy-Guided Test-Time Intervention for Large Language Model Reasoning","ref_index":9,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KPASI7VIC6IYHQLPWCAXGHSEUC","json":"https://pith.science/pith/KPASI7VIC6IYHQLPWCAXGHSEUC.json","graph_json":"https://pith.science/api/pith-number/KPASI7VIC6IYHQLPWCAXGHSEUC/graph.json","events_json":"https://pith.science/api/pith-number/KPASI7VIC6IYHQLPWCAXGHSEUC/events.json","paper":"https://pith.science/paper/KPASI7VI"},"agent_actions":{"view_html":"https://pith.science/pith/KPASI7VIC6IYHQLPWCAXGHSEUC","download_json":"https://pith.science/pith/KPASI7VIC6IYHQLPWCAXGHSEUC.json","view_paper":"https://pith.science/paper/KPASI7VI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2404.07965&json=true","fetch_graph":"https://pith.science/api/pith-number/KPASI7VIC6IYHQLPWCAXGHSEUC/graph.json","fetch_events":"https://pith.science/api/pith-number/KPASI7VIC6IYHQLPWCAXGHSEUC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KPASI7VIC6IYHQLPWCAXGHSEUC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KPASI7VIC6IYHQLPWCAXGHSEUC/action/storage_attestation","attest_author":"https://pith.science/pith/KPASI7VIC6IYHQLPWCAXGHSEUC/action/author_attestation","sign_citation":"https://pith.science/pith/KPASI7VIC6IYHQLPWCAXGHSEUC/action/citation_signature","submit_replication":"https://pith.science/pith/KPASI7VIC6IYHQLPWCAXGHSEUC/action/replication_record"}},"created_at":"2026-07-05T09:58:19.860308+00:00","updated_at":"2026-07-05T09:58:19.860308+00:00"}