{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:J2ZP6W3C32VB7BE6W7PY6LNXBR","short_pith_number":"pith:J2ZP6W3C","schema_version":"1.0","canonical_sha256":"4eb2ff5b62deaa1f849eb7df8f2db70c4f63c4ea4d62d18b0ef415a91440864b","source":{"kind":"arxiv","id":"2406.08464","version":2},"attestation_state":"computed","paper":{"title":"Magpie: Alignment Data Synthesis from Scratch by Prompting Aligned LLMs with Nothing","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Prompting aligned LLMs like Llama-3-Instruct with only left-side conversation templates produces millions of realistic user queries and responses for alignment training.","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Bill Yuchen Lin, Fengqing Jiang, Luyao Niu, Radha Poovendran, Yejin Choi, Yuntian Deng, Zhangchen Xu","submitted_at":"2024-06-12T17:52:30Z","abstract_excerpt":"High-quality instruction data is critical for aligning large language models (LLMs). Although some models, such as Llama-3-Instruct, have open weights, their alignment data remain private, which hinders the democratization of AI. High human labor costs and a limited, predefined scope for prompting prevent existing open-source data creation methods from scaling effectively, potentially limiting the diversity and quality of public alignment datasets. Is it possible to synthesize high-quality instruction data at scale by extracting it directly from an aligned LLM? We present a self-synthesis meth"},"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":true,"formal_links_present":false},"canonical_record":{"source":{"id":"2406.08464","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-12T17:52:30Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"aa1bd039007ba2f07d8106782e79a7b8ca176dfec737130f7a637de5d9eec978","abstract_canon_sha256":"aa7a052570c78eae2856548d7696d715493bb9648d3061b74e968934a92de08e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:48.766411Z","signature_b64":"bD1g2tiz6AgcGDemT3E2zmz6ReH8jyl3SfPor8Gtmc/63liemx73pz6L2BMvfe3PiVVL8YtW/BtnBwO+lyjhAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4eb2ff5b62deaa1f849eb7df8f2db70c4f63c4ea4d62d18b0ef415a91440864b","last_reissued_at":"2026-05-17T23:38:48.765889Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:48.765889Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Magpie: Alignment Data Synthesis from Scratch by Prompting Aligned LLMs with Nothing","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Prompting aligned LLMs like Llama-3-Instruct with only left-side conversation templates produces millions of realistic user queries and responses for alignment training.","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Bill Yuchen Lin, Fengqing Jiang, Luyao Niu, Radha Poovendran, Yejin Choi, Yuntian Deng, Zhangchen Xu","submitted_at":"2024-06-12T17:52:30Z","abstract_excerpt":"High-quality instruction data is critical for aligning large language models (LLMs). Although some models, such as Llama-3-Instruct, have open weights, their alignment data remain private, which hinders the democratization of AI. High human labor costs and a limited, predefined scope for prompting prevent existing open-source data creation methods from scaling effectively, potentially limiting the diversity and quality of public alignment datasets. Is it possible to synthesize high-quality instruction data at scale by extracting it directly from an aligned LLM? We present a self-synthesis meth"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Our results indicate that in some tasks, models fine-tuned with Magpie perform comparably to the official Llama-3-8B-Instruct, despite the latter being enhanced with 10 million data points through supervised fine-tuning (SFT) and subsequent feedback learning.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The generated user queries produced by prompting with left-side templates are sufficiently diverse, realistic, and representative of real user needs to support effective alignment after filtering.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Magpie synthesizes 300K high-quality alignment instructions from Llama-3-Instruct via auto-regressive prompting on partial templates, enabling fine-tuned models to match official instruct performance on AlpacaEval, ArenaHard, and WildBench.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Prompting aligned LLMs like Llama-3-Instruct with only left-side conversation templates produces millions of realistic user queries and responses for alignment training.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"da8a76606000b613e379f3e979ba4b857811040014e2997dc7d20ee7c283bb73"},"source":{"id":"2406.08464","kind":"arxiv","version":2},"verdict":{"id":"4a54f3eb-5e78-4e05-9cc7-81fe03eab948","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T06:54:57.162248Z","strongest_claim":"Our results indicate that in some tasks, models fine-tuned with Magpie perform comparably to the official Llama-3-8B-Instruct, despite the latter being enhanced with 10 million data points through supervised fine-tuning (SFT) and subsequent feedback learning.","one_line_summary":"Magpie synthesizes 300K high-quality alignment instructions from Llama-3-Instruct via auto-regressive prompting on partial templates, enabling fine-tuned models to match official instruct performance on AlpacaEval, ArenaHard, and WildBench.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The generated user queries produced by prompting with left-side templates are sufficiently diverse, realistic, and representative of real user needs to support effective alignment after filtering.","pith_extraction_headline":"Prompting aligned LLMs like Llama-3-Instruct with only left-side conversation templates produces millions of realistic user queries and responses for alignment training."},"references":{"count":126,"sample":[{"doi":"","year":null,"title":"and Stoica, Ion and Xing, Eric P","work_id":"cb4b41f6-6d60-4db4-a4d1-6c5bb7899473","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Interpretable Preferences via Multi-Objective Reward Modeling and Mixture-of-Experts , author=. EMNLP , year=","work_id":"77415ed4-7f1f-44b1-bd55-58146dac1904","ref_index":6,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency , pages=","work_id":"bd104d56-1e1d-4622-9852-13b7c597c1ae","ref_index":8,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2018,"title":"Improving language understanding by generative pre-training , author=. 2018 , publisher=","work_id":"3bddebd1-6efa-4e33-b5e0-8e75c670b486","ref_index":9,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"International Conference on Machine Learning , pages=","work_id":"e6670a94-9685-42c2-84ec-6accc28bc8a4","ref_index":11,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":126,"snapshot_sha256":"f93013ac98f5fddef0513f65b115c13598a8e648edab6b1abca9114f378a3901","internal_anchors":22},"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":"2406.08464","created_at":"2026-05-17T23:38:48.765982+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.08464v2","created_at":"2026-05-17T23:38:48.765982+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.08464","created_at":"2026-05-17T23:38:48.765982+00:00"},{"alias_kind":"pith_short_12","alias_value":"J2ZP6W3C32VB","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"J2ZP6W3C32VB7BE6","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"J2ZP6W3C","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":28,"internal_anchor_count":28,"sample":[{"citing_arxiv_id":"2406.11354","citing_title":"Preserving Knowledge in Large Language Model with Model-Agnostic Self-Decompression","ref_index":60,"is_internal_anchor":true},{"citing_arxiv_id":"2502.05206","citing_title":"Safety at Scale: A Comprehensive Survey of Large Model and Agent Safety","ref_index":209,"is_internal_anchor":true},{"citing_arxiv_id":"2509.03526","citing_title":"Enhancing Speech Large Language Models through Reinforced Behavior Alignment","ref_index":52,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14381","citing_title":"NodeSynth: Socially Aligned Synthetic Data for AI Evaluation","ref_index":3,"is_internal_anchor":true},{"citing_arxiv_id":"2506.15155","citing_title":"eLLM: Elastic Memory Management Framework for Efficient LLM Serving","ref_index":36,"is_internal_anchor":true},{"citing_arxiv_id":"2507.21166","citing_title":"The Ratchet Effect in Silico through Interaction-Driven Cumulative Intelligence in Large Language Models","ref_index":51,"is_internal_anchor":true},{"citing_arxiv_id":"2410.18451","citing_title":"Skywork-Reward: Bag of Tricks for Reward Modeling in LLMs","ref_index":23,"is_internal_anchor":true},{"citing_arxiv_id":"2602.06019","citing_title":"Multi-Token Prediction via Self-Distillation","ref_index":18,"is_internal_anchor":true},{"citing_arxiv_id":"2604.09557","citing_title":"SPEED-Bench: A Unified and Diverse Benchmark for Speculative Decoding","ref_index":53,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14381","citing_title":"NodeSynth: Socially Aligned Synthetic Data for AI Evaluation","ref_index":3,"is_internal_anchor":true},{"citing_arxiv_id":"2605.11679","citing_title":"Explaining and Breaking the Safety-Helpfulness Ceiling via Preference Dimensional Expansion","ref_index":60,"is_internal_anchor":true},{"citing_arxiv_id":"2504.05299","citing_title":"SmolVLM: Redefining small and efficient multimodal models","ref_index":34,"is_internal_anchor":true},{"citing_arxiv_id":"2502.02737","citing_title":"SmolLM2: When Smol Goes Big -- Data-Centric Training of a Small Language Model","ref_index":242,"is_internal_anchor":true},{"citing_arxiv_id":"2605.11679","citing_title":"Explaining and Breaking the Safety-Helpfulness Ceiling via Preference Dimensional Expansion","ref_index":60,"is_internal_anchor":true},{"citing_arxiv_id":"2502.05171","citing_title":"Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach","ref_index":171,"is_internal_anchor":true},{"citing_arxiv_id":"2604.27861","citing_title":"TwinGate: Stateful Defense against Decompositional Jailbreaks in Untraceable Traffic via Asymmetric Contrastive Learning","ref_index":32,"is_internal_anchor":true},{"citing_arxiv_id":"2605.08632","citing_title":"PARD-2: Target-Aligned Parallel Draft Model for Dual-Mode Speculative Decoding","ref_index":31,"is_internal_anchor":true},{"citing_arxiv_id":"2504.18425","citing_title":"Kimi-Audio Technical Report","ref_index":75,"is_internal_anchor":true},{"citing_arxiv_id":"2605.04913","citing_title":"Rethinking Local Learning: A Cheaper and Faster Recipe for LLM Post-Training","ref_index":27,"is_internal_anchor":true},{"citing_arxiv_id":"2605.05365","citing_title":"ZAYA1-8B Technical Report","ref_index":65,"is_internal_anchor":true},{"citing_arxiv_id":"2412.10302","citing_title":"DeepSeek-VL2: Mixture-of-Experts Vision-Language Models for Advanced Multimodal Understanding","ref_index":99,"is_internal_anchor":true},{"citing_arxiv_id":"2605.04913","citing_title":"Rethinking Local Learning: A Cheaper and Faster Recipe for LLM Post-Training","ref_index":27,"is_internal_anchor":true},{"citing_arxiv_id":"2502.09992","citing_title":"Large Language Diffusion Models","ref_index":104,"is_internal_anchor":true},{"citing_arxiv_id":"2501.13106","citing_title":"VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understanding","ref_index":95,"is_internal_anchor":true},{"citing_arxiv_id":"2604.06854","citing_title":"To Adapt or not to Adapt, Rethinking the Value of Medical Knowledge-Aware Large Language Models","ref_index":5,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/J2ZP6W3C32VB7BE6W7PY6LNXBR","json":"https://pith.science/pith/J2ZP6W3C32VB7BE6W7PY6LNXBR.json","graph_json":"https://pith.science/api/pith-number/J2ZP6W3C32VB7BE6W7PY6LNXBR/graph.json","events_json":"https://pith.science/api/pith-number/J2ZP6W3C32VB7BE6W7PY6LNXBR/events.json","paper":"https://pith.science/paper/J2ZP6W3C"},"agent_actions":{"view_html":"https://pith.science/pith/J2ZP6W3C32VB7BE6W7PY6LNXBR","download_json":"https://pith.science/pith/J2ZP6W3C32VB7BE6W7PY6LNXBR.json","view_paper":"https://pith.science/paper/J2ZP6W3C","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.08464&json=true","fetch_graph":"https://pith.science/api/pith-number/J2ZP6W3C32VB7BE6W7PY6LNXBR/graph.json","fetch_events":"https://pith.science/api/pith-number/J2ZP6W3C32VB7BE6W7PY6LNXBR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/J2ZP6W3C32VB7BE6W7PY6LNXBR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/J2ZP6W3C32VB7BE6W7PY6LNXBR/action/storage_attestation","attest_author":"https://pith.science/pith/J2ZP6W3C32VB7BE6W7PY6LNXBR/action/author_attestation","sign_citation":"https://pith.science/pith/J2ZP6W3C32VB7BE6W7PY6LNXBR/action/citation_signature","submit_replication":"https://pith.science/pith/J2ZP6W3C32VB7BE6W7PY6LNXBR/action/replication_record"}},"created_at":"2026-05-17T23:38:48.765982+00:00","updated_at":"2026-05-17T23:38:48.765982+00:00"}