{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:J2ZP6W3C32VB7BE6W7PY6LNXBR","short_pith_number":"pith:J2ZP6W3C","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"},"canonical_sha256":"4eb2ff5b62deaa1f849eb7df8f2db70c4f63c4ea4d62d18b0ef415a91440864b","source":{"kind":"arxiv","id":"2406.08464","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.08464","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"arxiv_version","alias_value":"2406.08464v2","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.08464","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"pith_short_12","alias_value":"J2ZP6W3C32VB","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"J2ZP6W3C32VB7BE6","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"J2ZP6W3C","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:J2ZP6W3C32VB7BE6W7PY6LNXBR","target":"record","payload":{"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"},"canonical_sha256":"4eb2ff5b62deaa1f849eb7df8f2db70c4f63c4ea4d62d18b0ef415a91440864b","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"},"source_kind":"arxiv","source_id":"2406.08464","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"4Hj+sVQegjjkayIYgZUBlUtrhGZD10eFfkC7sHskdek1QgiprLdCZ9DRdSIax6+I5X5RyGfMjQ/h0GxSgg+bAQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-30T13:26:15.842692Z"},"content_sha256":"9638b7d775ff0364ac5bac3cd33afd54b909b28cb7c082c29381525068c073b1","schema_version":"1.0","event_id":"sha256:9638b7d775ff0364ac5bac3cd33afd54b909b28cb7c082c29381525068c073b1"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:J2ZP6W3C32VB7BE6W7PY6LNXBR","target":"graph","payload":{"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"},"verdict_id":"4a54f3eb-5e78-4e05-9cc7-81fe03eab948"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:48Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"oLNSfcucJImNEo8EAe/GUSdPnGrqKstrLQDBcfuJkUPgLEOQYtJpvsEIhK5VdDjTu0aOJBv4/eJew0yNktb7Dw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-30T13:26:15.843226Z"},"content_sha256":"2d6666bad114626d9805cf6231a138bcd5a4fcb593f79f6a78d66aca54223fc4","schema_version":"1.0","event_id":"sha256:2d6666bad114626d9805cf6231a138bcd5a4fcb593f79f6a78d66aca54223fc4"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/J2ZP6W3C32VB7BE6W7PY6LNXBR/bundle.json","state_url":"https://pith.science/pith/J2ZP6W3C32VB7BE6W7PY6LNXBR/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/J2ZP6W3C32VB7BE6W7PY6LNXBR/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-05-30T13:26:15Z","links":{"resolver":"https://pith.science/pith/J2ZP6W3C32VB7BE6W7PY6LNXBR","bundle":"https://pith.science/pith/J2ZP6W3C32VB7BE6W7PY6LNXBR/bundle.json","state":"https://pith.science/pith/J2ZP6W3C32VB7BE6W7PY6LNXBR/state.json","well_known_bundle":"https://pith.science/.well-known/pith/J2ZP6W3C32VB7BE6W7PY6LNXBR/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:J2ZP6W3C32VB7BE6W7PY6LNXBR","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"aa7a052570c78eae2856548d7696d715493bb9648d3061b74e968934a92de08e","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-12T17:52:30Z","title_canon_sha256":"aa1bd039007ba2f07d8106782e79a7b8ca176dfec737130f7a637de5d9eec978"},"schema_version":"1.0","source":{"id":"2406.08464","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.08464","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"arxiv_version","alias_value":"2406.08464v2","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.08464","created_at":"2026-05-17T23:38:48Z"},{"alias_kind":"pith_short_12","alias_value":"J2ZP6W3C32VB","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"J2ZP6W3C32VB7BE6","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"J2ZP6W3C","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:2d6666bad114626d9805cf6231a138bcd5a4fcb593f79f6a78d66aca54223fc4","target":"graph","created_at":"2026-05-17T23:38:48Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","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."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","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."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","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."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","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."}],"snapshot_sha256":"da8a76606000b613e379f3e979ba4b857811040014e2997dc7d20ee7c283bb73"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"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","authors_text":"Bill Yuchen Lin, Fengqing Jiang, Luyao Niu, Radha Poovendran, Yejin Choi, Yuntian Deng, Zhangchen Xu","cross_cats":["cs.AI"],"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.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-12T17:52:30Z","title":"Magpie: Alignment Data Synthesis from Scratch by Prompting Aligned LLMs with Nothing"},"references":{"count":126,"internal_anchors":22,"resolved_work":126,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"and Stoica, Ion and Xing, Eric P","work_id":"cb4b41f6-6d60-4db4-a4d1-6c5bb7899473","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":6,"title":"Interpretable Preferences via Multi-Objective Reward Modeling and Mixture-of-Experts , author=. EMNLP , year=","work_id":"77415ed4-7f1f-44b1-bd55-58146dac1904","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":8,"title":"Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency , pages=","work_id":"bd104d56-1e1d-4622-9852-13b7c597c1ae","year":2022},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":9,"title":"Improving language understanding by generative pre-training , author=. 2018 , publisher=","work_id":"3bddebd1-6efa-4e33-b5e0-8e75c670b486","year":2018},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":11,"title":"International Conference on Machine Learning , pages=","work_id":"e6670a94-9685-42c2-84ec-6accc28bc8a4","year":2023}],"snapshot_sha256":"f93013ac98f5fddef0513f65b115c13598a8e648edab6b1abca9114f378a3901"},"source":{"id":"2406.08464","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-16T06:54:57.162248Z","id":"4a54f3eb-5e78-4e05-9cc7-81fe03eab948","model_set":{"reader":"grok-4.3"},"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","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.","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.","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."}},"verdict_id":"4a54f3eb-5e78-4e05-9cc7-81fe03eab948"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:9638b7d775ff0364ac5bac3cd33afd54b909b28cb7c082c29381525068c073b1","target":"record","created_at":"2026-05-17T23:38:48Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"aa7a052570c78eae2856548d7696d715493bb9648d3061b74e968934a92de08e","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-12T17:52:30Z","title_canon_sha256":"aa1bd039007ba2f07d8106782e79a7b8ca176dfec737130f7a637de5d9eec978"},"schema_version":"1.0","source":{"id":"2406.08464","kind":"arxiv","version":2}},"canonical_sha256":"4eb2ff5b62deaa1f849eb7df8f2db70c4f63c4ea4d62d18b0ef415a91440864b","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"4eb2ff5b62deaa1f849eb7df8f2db70c4f63c4ea4d62d18b0ef415a91440864b","first_computed_at":"2026-05-17T23:38:48.765889Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:48.765889Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"bD1g2tiz6AgcGDemT3E2zmz6ReH8jyl3SfPor8Gtmc/63liemx73pz6L2BMvfe3PiVVL8YtW/BtnBwO+lyjhAA==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:48.766411Z","signed_message":"canonical_sha256_bytes"},"source_id":"2406.08464","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:9638b7d775ff0364ac5bac3cd33afd54b909b28cb7c082c29381525068c073b1","sha256:2d6666bad114626d9805cf6231a138bcd5a4fcb593f79f6a78d66aca54223fc4"],"state_sha256":"6f68c24df351214f293417c53e4e5f0b126d1ddb583a22f96aa8c6fae8c35aa9"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"/d4EEYSiTyfBhkWsq+8l2h/Hadna5KRLJ95dxfBQix2Xe14CYZaAUQ3kYmMz0aFFLZVLFrZUujHc2POyq1CZBw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-30T13:26:15.845647Z","bundle_sha256":"5af4c6687757a6e3f485dde3d10039552aad5e8019b9076da456eec096e6b716"}}