{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:YKCLRBMGAWNJATFOD2E5WVPH5J","short_pith_number":"pith:YKCLRBMG","canonical_record":{"source":{"id":"2410.10762","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2024-10-14T17:40:40Z","cross_cats_sorted":["cs.CL","cs.LG","cs.SE"],"title_canon_sha256":"6eb98f7dbf700f1f4bf1aa30b3fe0fd739e9e61c275405e38de7f12aaab23c98","abstract_canon_sha256":"72af6de20abb7dc257f29c7597c88f76e2bd17da4d30c49ebc056dedd4e90844"},"schema_version":"1.0"},"canonical_sha256":"c284b88586059a904cae1e89db55e7ea7fdaebbf020f5c448eda793350db32f2","source":{"kind":"arxiv","id":"2410.10762","version":4},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2410.10762","created_at":"2026-05-17T23:38:53Z"},{"alias_kind":"arxiv_version","alias_value":"2410.10762v4","created_at":"2026-05-17T23:38:53Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.10762","created_at":"2026-05-17T23:38:53Z"},{"alias_kind":"pith_short_12","alias_value":"YKCLRBMGAWNJ","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"YKCLRBMGAWNJATFO","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"YKCLRBMG","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:YKCLRBMGAWNJATFOD2E5WVPH5J","target":"record","payload":{"canonical_record":{"source":{"id":"2410.10762","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2024-10-14T17:40:40Z","cross_cats_sorted":["cs.CL","cs.LG","cs.SE"],"title_canon_sha256":"6eb98f7dbf700f1f4bf1aa30b3fe0fd739e9e61c275405e38de7f12aaab23c98","abstract_canon_sha256":"72af6de20abb7dc257f29c7597c88f76e2bd17da4d30c49ebc056dedd4e90844"},"schema_version":"1.0"},"canonical_sha256":"c284b88586059a904cae1e89db55e7ea7fdaebbf020f5c448eda793350db32f2","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:53.711486Z","signature_b64":"eiLIBpMhCEicThvW1QXykLWXT0FMqUSIFaBaL6c2C5gpEfcppcuLyjlPpEkQU4O2mSUdFTXlgWLswgJq7gxkCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c284b88586059a904cae1e89db55e7ea7fdaebbf020f5c448eda793350db32f2","last_reissued_at":"2026-05-17T23:38:53.710970Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:53.710970Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2410.10762","source_version":4,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:53Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"YB+9ejQDj1H00OnKoXHGkIn6sujlu3smMhAY6yA9g26y/L7Gel7V74f/J1a+Cha7JIP8dtKvmjyhAZG3iliIBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-04T20:26:30.836589Z"},"content_sha256":"ef227f65d73c49acc596e2e81d14e158f1cda507c8e4db589d8cbb0f63a0c67d","schema_version":"1.0","event_id":"sha256:ef227f65d73c49acc596e2e81d14e158f1cda507c8e4db589d8cbb0f63a0c67d"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:YKCLRBMGAWNJATFOD2E5WVPH5J","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"AFlow: Automating Agentic Workflow Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Code search automates LLM workflows with 5.7% performance gains","cross_cats":["cs.CL","cs.LG","cs.SE"],"primary_cat":"cs.AI","authors_text":"Bang Liu, Bingnan Zheng, Chenglin Wu, Fengwei Teng, Jiaqi Chen, Jiayi Zhang, Jinlin Wang, Jinyu Xiang, Mingchen Zhuge, Sirui Hong, Xin Cheng, Xionghui Chen, Yuyu Luo, Zhaoyang Yu","submitted_at":"2024-10-14T17:40:40Z","abstract_excerpt":"Large language models (LLMs) have demonstrated remarkable potential in solving complex tasks across diverse domains, typically by employing agentic workflows that follow detailed instructions and operational sequences. However, constructing these workflows requires significant human effort, limiting scalability and generalizability. Recent research has sought to automate the generation and optimization of these workflows, but existing methods still rely on initial manual setup and fall short of achieving fully automated and effective workflow generation. To address this challenge, we reformula"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Empirical evaluations across six benchmark datasets demonstrate AFlow's efficacy, yielding a 5.7% average improvement over state-of-the-art baselines. Furthermore, AFlow enables smaller models to outperform GPT-4o on specific tasks at 4.55% of its inference cost in dollars.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the space of code-represented workflows can be searched efficiently by Monte Carlo Tree Search with code edits and execution feedback without excessive compute or getting trapped in poor local solutions.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"AFlow uses Monte Carlo Tree Search to automatically generate and optimize code-represented agentic workflows for LLMs, delivering a 5.7% average gain over prior methods on six benchmarks while letting smaller models beat GPT-4o at low cost.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Code search automates LLM workflows with 5.7% performance gains","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"0aaf50ab712c6c228e94e46992b94fd2887474d1e05ade4ddcd7ce47e5ba4d57"},"source":{"id":"2410.10762","kind":"arxiv","version":4},"verdict":{"id":"e6ec8376-b1fd-45c9-9091-7193e3d29296","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T03:04:59.474821Z","strongest_claim":"Empirical evaluations across six benchmark datasets demonstrate AFlow's efficacy, yielding a 5.7% average improvement over state-of-the-art baselines. Furthermore, AFlow enables smaller models to outperform GPT-4o on specific tasks at 4.55% of its inference cost in dollars.","one_line_summary":"AFlow uses Monte Carlo Tree Search to automatically generate and optimize code-represented agentic workflows for LLMs, delivering a 5.7% average gain over prior methods on six benchmarks while letting smaller models beat GPT-4o at low cost.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the space of code-represented workflows can be searched efficiently by Monte Carlo Tree Search with code edits and execution feedback without excessive compute or getting trapped in poor local solutions.","pith_extraction_headline":"Code search automates LLM workflows with 5.7% performance gains"},"references":{"count":63,"sample":[{"doi":"","year":null,"title":"Begin with a clear statement of the problem","work_id":"88d279aa-f22f-4edd-afab-f3757a169a9d","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Explain the approach and any formulas or concepts used","work_id":"3eed3c13-e959-4090-bb89-40335b4408af","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Show step-by-step calculations, using LaTeX notation for mathematical expressions","work_id":"a5391493-0aa2-46f3-b9c6-ee1c779ba31f","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Interpret the code output and incorporate it into your explanation","work_id":"70551d31-5799-4a4d-bbf8-a07e0d209706","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Provide a final answer, enclosed in \\boxed{} LaTeX notation","work_id":"ce8747c0-b767-4176-8d73-1d34120ec720","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":63,"snapshot_sha256":"25618a8bdc6c9d4d2ee00612c9913f454216a684cc75959cbf7224a8e9ae831c","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"},"verdict_id":"e6ec8376-b1fd-45c9-9091-7193e3d29296"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:53Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"joQwT7Zg3/HBW/kotbfeadHmZAPtsgqaTvtNR95zbOLnzf04E9cGzbOz9JVevRHVHCjNN+Enr44wiaOLmMQ+Dw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-04T20:26:30.837125Z"},"content_sha256":"0f5ccb9334d5db749a13314538fe99a19a89f5d5e2d0ec85824f16bd4a2addd7","schema_version":"1.0","event_id":"sha256:0f5ccb9334d5db749a13314538fe99a19a89f5d5e2d0ec85824f16bd4a2addd7"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/YKCLRBMGAWNJATFOD2E5WVPH5J/bundle.json","state_url":"https://pith.science/pith/YKCLRBMGAWNJATFOD2E5WVPH5J/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/YKCLRBMGAWNJATFOD2E5WVPH5J/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":"2026-06-04T20:26:30Z","links":{"resolver":"https://pith.science/pith/YKCLRBMGAWNJATFOD2E5WVPH5J","bundle":"https://pith.science/pith/YKCLRBMGAWNJATFOD2E5WVPH5J/bundle.json","state":"https://pith.science/pith/YKCLRBMGAWNJATFOD2E5WVPH5J/state.json","well_known_bundle":"https://pith.science/.well-known/pith/YKCLRBMGAWNJATFOD2E5WVPH5J/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:YKCLRBMGAWNJATFOD2E5WVPH5J","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":"72af6de20abb7dc257f29c7597c88f76e2bd17da4d30c49ebc056dedd4e90844","cross_cats_sorted":["cs.CL","cs.LG","cs.SE"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2024-10-14T17:40:40Z","title_canon_sha256":"6eb98f7dbf700f1f4bf1aa30b3fe0fd739e9e61c275405e38de7f12aaab23c98"},"schema_version":"1.0","source":{"id":"2410.10762","kind":"arxiv","version":4}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2410.10762","created_at":"2026-05-17T23:38:53Z"},{"alias_kind":"arxiv_version","alias_value":"2410.10762v4","created_at":"2026-05-17T23:38:53Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.10762","created_at":"2026-05-17T23:38:53Z"},{"alias_kind":"pith_short_12","alias_value":"YKCLRBMGAWNJ","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"YKCLRBMGAWNJATFO","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"YKCLRBMG","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:0f5ccb9334d5db749a13314538fe99a19a89f5d5e2d0ec85824f16bd4a2addd7","target":"graph","created_at":"2026-05-17T23:38:53Z","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":"Empirical evaluations across six benchmark datasets demonstrate AFlow's efficacy, yielding a 5.7% average improvement over state-of-the-art baselines. Furthermore, AFlow enables smaller models to outperform GPT-4o on specific tasks at 4.55% of its inference cost in dollars."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the space of code-represented workflows can be searched efficiently by Monte Carlo Tree Search with code edits and execution feedback without excessive compute or getting trapped in poor local solutions."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"AFlow uses Monte Carlo Tree Search to automatically generate and optimize code-represented agentic workflows for LLMs, delivering a 5.7% average gain over prior methods on six benchmarks while letting smaller models beat GPT-4o at low cost."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Code search automates LLM workflows with 5.7% performance gains"}],"snapshot_sha256":"0aaf50ab712c6c228e94e46992b94fd2887474d1e05ade4ddcd7ce47e5ba4d57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"Large language models (LLMs) have demonstrated remarkable potential in solving complex tasks across diverse domains, typically by employing agentic workflows that follow detailed instructions and operational sequences. However, constructing these workflows requires significant human effort, limiting scalability and generalizability. Recent research has sought to automate the generation and optimization of these workflows, but existing methods still rely on initial manual setup and fall short of achieving fully automated and effective workflow generation. To address this challenge, we reformula","authors_text":"Bang Liu, Bingnan Zheng, Chenglin Wu, Fengwei Teng, Jiaqi Chen, Jiayi Zhang, Jinlin Wang, Jinyu Xiang, Mingchen Zhuge, Sirui Hong, Xin Cheng, Xionghui Chen, Yuyu Luo, Zhaoyang Yu","cross_cats":["cs.CL","cs.LG","cs.SE"],"headline":"Code search automates LLM workflows with 5.7% performance gains","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2024-10-14T17:40:40Z","title":"AFlow: Automating Agentic Workflow Generation"},"references":{"count":63,"internal_anchors":0,"resolved_work":63,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Begin with a clear statement of the problem","work_id":"88d279aa-f22f-4edd-afab-f3757a169a9d","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Explain the approach and any formulas or concepts used","work_id":"3eed3c13-e959-4090-bb89-40335b4408af","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Show step-by-step calculations, using LaTeX notation for mathematical expressions","work_id":"a5391493-0aa2-46f3-b9c6-ee1c779ba31f","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Interpret the code output and incorporate it into your explanation","work_id":"70551d31-5799-4a4d-bbf8-a07e0d209706","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Provide a final answer, enclosed in \\boxed{} LaTeX notation","work_id":"ce8747c0-b767-4176-8d73-1d34120ec720","year":null}],"snapshot_sha256":"25618a8bdc6c9d4d2ee00612c9913f454216a684cc75959cbf7224a8e9ae831c"},"source":{"id":"2410.10762","kind":"arxiv","version":4},"verdict":{"created_at":"2026-05-15T03:04:59.474821Z","id":"e6ec8376-b1fd-45c9-9091-7193e3d29296","model_set":{"reader":"grok-4.3"},"one_line_summary":"AFlow uses Monte Carlo Tree Search to automatically generate and optimize code-represented agentic workflows for LLMs, delivering a 5.7% average gain over prior methods on six benchmarks while letting smaller models beat GPT-4o at low cost.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Code search automates LLM workflows with 5.7% performance gains","strongest_claim":"Empirical evaluations across six benchmark datasets demonstrate AFlow's efficacy, yielding a 5.7% average improvement over state-of-the-art baselines. Furthermore, AFlow enables smaller models to outperform GPT-4o on specific tasks at 4.55% of its inference cost in dollars.","weakest_assumption":"That the space of code-represented workflows can be searched efficiently by Monte Carlo Tree Search with code edits and execution feedback without excessive compute or getting trapped in poor local solutions."}},"verdict_id":"e6ec8376-b1fd-45c9-9091-7193e3d29296"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:ef227f65d73c49acc596e2e81d14e158f1cda507c8e4db589d8cbb0f63a0c67d","target":"record","created_at":"2026-05-17T23:38:53Z","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":"72af6de20abb7dc257f29c7597c88f76e2bd17da4d30c49ebc056dedd4e90844","cross_cats_sorted":["cs.CL","cs.LG","cs.SE"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2024-10-14T17:40:40Z","title_canon_sha256":"6eb98f7dbf700f1f4bf1aa30b3fe0fd739e9e61c275405e38de7f12aaab23c98"},"schema_version":"1.0","source":{"id":"2410.10762","kind":"arxiv","version":4}},"canonical_sha256":"c284b88586059a904cae1e89db55e7ea7fdaebbf020f5c448eda793350db32f2","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"c284b88586059a904cae1e89db55e7ea7fdaebbf020f5c448eda793350db32f2","first_computed_at":"2026-05-17T23:38:53.710970Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:53.710970Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"eiLIBpMhCEicThvW1QXykLWXT0FMqUSIFaBaL6c2C5gpEfcppcuLyjlPpEkQU4O2mSUdFTXlgWLswgJq7gxkCw==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:53.711486Z","signed_message":"canonical_sha256_bytes"},"source_id":"2410.10762","source_kind":"arxiv","source_version":4}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:ef227f65d73c49acc596e2e81d14e158f1cda507c8e4db589d8cbb0f63a0c67d","sha256:0f5ccb9334d5db749a13314538fe99a19a89f5d5e2d0ec85824f16bd4a2addd7"],"state_sha256":"feee157d466757f6ef0c1ca5a07e72d3a4597aa08c68998be68ae63c89c207b7"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"o/DjS+DCngJgnZXKNmUbXC8GfEXYxlnFa+HihE/IJbyoZ+XbRAxRCX5XNa8eF0uObZguIHrI5ApAprt+Gw5pBg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-04T20:26:30.839526Z","bundle_sha256":"dfe90ba3bc624259f1c780ebba58c445f4d61bb367e01eba3f30b71d17372e19"}}