{"work":{"id":"0e0b7549-2bc4-4574-aa7f-588ffa16eaae","openalex_id":null,"doi":null,"arxiv_id":"2503.09516","raw_key":null,"title":"Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning","authors":null,"authors_text":"Bowen Jin, Hansi Zeng, Zhenrui Yue, Jinsung Yoon, Sercan Arik, Dong Wang","year":2025,"venue":"cs.CL","abstract":"Efficiently acquiring external knowledge and up-to-date information is essential for effective reasoning and text generation in large language models (LLMs). Prompting advanced LLMs with reasoning capabilities to use search engines during inference is often suboptimal, as the LLM might not fully possess the capability on how to interact optimally with the search engine. This paper introduces Search-R1, an extension of reinforcement learning (RL) for reasoning frameworks where the LLM learns to autonomously generate (multiple) search queries during step-by-step reasoning with real-time retrieval. Search-R1 optimizes LLM reasoning trajectories with multi-turn search interactions, leveraging retrieved token masking for stable RL training and a simple outcome-based reward function. Experiments on seven question-answering datasets show that Search-R1 improves performance by 41% (Qwen2.5-7B) and 20% (Qwen2.5-3B) over various RAG baselines under the same setting. This paper further provides empirical insights into RL optimization methods, LLM choices, and response length dynamics in retrieval-augmented reasoning. The code and model checkpoints are available at https://github.com/PeterGriffinJin/Search-R1.","external_url":"https://arxiv.org/abs/2503.09516","cited_by_count":null,"metadata_source":"pith","metadata_fetched_at":"2026-05-25T07:45:28.553524+00:00","pith_arxiv_id":"2503.09516","created_at":"2026-05-09T05:45:22.143841+00:00","updated_at":"2026-05-25T07:45:28.553524+00:00","title_quality_ok":true,"display_title":"Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning","render_title":"Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning"},"hub":{"state":{"work_id":"0e0b7549-2bc4-4574-aa7f-588ffa16eaae","tier":"super_hub","tier_reason":"100+ Pith inbound or 10,000+ external citations","pith_inbound_count":125,"external_cited_by_count":null,"distinct_field_count":7,"first_pith_cited_at":"2025-02-19T18:56:03+00:00","last_pith_cited_at":"2026-05-22T12:59:16+00:00","author_build_status":"needed","summary_status":"needed","contexts_status":"needed","graph_status":"needed","ask_index_status":"needed","reader_status":"not_needed","recognition_status":"not_needed","updated_at":"2026-06-02T00:34:01.010621+00:00","tier_text":"super_hub"},"tier":"super_hub","role_counts":[{"context_role":"background","n":42},{"context_role":"method","n":3},{"context_role":"baseline","n":2},{"context_role":"dataset","n":1},{"context_role":"extension","n":1},{"context_role":"other","n":1}],"polarity_counts":[{"context_polarity":"background","n":39},{"context_polarity":"unclear","n":4},{"context_polarity":"use_method","n":3},{"context_polarity":"baseline","n":2},{"context_polarity":"extend","n":1},{"context_polarity":"use_dataset","n":1}],"runs":{"ask_index":{"job_type":"ask_index","status":"succeeded","result":{"title":"Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning","claims":[{"claim_text":"Efficiently acquiring external knowledge and up-to-date information is essential for effective reasoning and text generation in large language models (LLMs). Prompting advanced LLMs with reasoning capabilities to use search engines during inference is often suboptimal, as the LLM might not fully possess the capability on how to interact optimally with the search engine. This paper introduces Search-R1, an extension of reinforcement learning (RL) for reasoning frameworks where the LLM learns to autonomously generate (multiple) search queries during step-by-step reasoning with real-time retrieva","claim_type":"abstract","evidence_strength":"source_metadata"},{"claim_text":"turn search horizons, achieving state-of-the-art performance among open-source models on challenging long-horizon benchmarks like MM-BrowseComp and MMSearch-Plus, while also exhibiting strong generalizability across different base models. Our code will be released in https://github.com/RUCAIBox/ LMM-Searcher. 1 Introduction Deep search agent systems [1, 2, 3] have achieved significant success in tackling challenging real- world information-seeking problems [4, 5]. Building on the deep search fra","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"org/abs/2402.01680. [9] Jie Huang and Kevin Chen-Chuan Chang. Towards reasoning in large language models: A survey. InFindings of the Association for Computational Linguistics: ACL 2023, pages 1049-1065, Toronto, Canada, 2023. Association for Computational Linguistics. doi: 10.18653/v1/2023.findings-acl.67. URL https://aclanthology.org/2023.findings-acl.67/. [10] Bowen Jin, Hansi Zeng, Zhenrui Yue, Jinsung Yoon, Sercan Arik, Dong Wang, Hamed Zamani, and Jiawei Han. Search-r1: Training llms to re","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"We evaluate three representative RL pipelines for agent training.Single-Turngenerates a complete response in one step, commonly used for reasoning tasks [16, 21]. Multi-Turniteratively produces intermediate outputs for self- refinement and multi-agent interaction [8, 25, 41, 53].Tool- Usageinvokes external tools during rollout; we use the Search-R1 pipeline [23]. We adopt GRPO [43] for policy optimization, with a global batch size of 64, group size 4, and maximum response length of 8192 tokens. ","claim_type":"method","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"Xu, Zhongyu Zhang, and Zhen Zhang. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning, 2025. URL https://arxiv.org/abs/2501.12948. [5] Luyu Gao, Aman Madaan, Shuyan Zhou, Uri Alon, Pengfei Liu, Yiming Yang, Jamie Callan, and Graham Neubig. Pal: Program-aided language models, 2023. URLhttps://arxiv.org/abs/2211.10435. [6] Bowen Jin, Hansi Zeng, Zhenrui Yue, Jinsung Yoon, Sercan Arik, Dong Wang, Hamed Zamani, and Jiawei Han. Search-r1: Training llms to reason and le","claim_type":"other","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"RL algorithms have emerged as promising alternatives to traditional methods like PPO [42]. These methods, such as RLOO [16, 17], GRPO [18], Dr. GRPO [19], DAPO [20], and CPPO [57], avoid introducing extra value functions by leveraging a group of samples from the same query and estimate the advantages accordingly. This enables the large-scale RL training and has shown strong results in tasks such as mathematical reasoning [15], search [58, 59], and tool use [60, 61]. Our work is closely related t","claim_type":"baseline","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"ods commonly adopt a translation-based strategy [1, 2, 13, 28, 31], translating documents into a single language before retrieval and generation. This approach reduces language gaps to some extent, but noise introduced by translation errors may lead to the loss of key information. Other methods focus on modifying the knowl- edge injection process [8, 41] or the reasoning strategy [16, 19, 32]. However, modifying knowledge injection often disrupts semantic integrity, while complex reasoning strat","claim_type":"background","confidence":0.85,"evidence_strength":"citation_context"}],"why_cited":"Pith tracks Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning because it crossed a citation-hub threshold. Current citing contexts most often use it as background evidence (23 contexts).","role_counts":[{"n":23,"context_role":"background"},{"n":3,"context_role":"method"},{"n":1,"context_role":"baseline"},{"n":1,"context_role":"other"}]},"error":null,"updated_at":"2026-05-18T09:30:39.973200+00:00"},"author_expand":{"job_type":"author_expand","status":"succeeded","result":{"authors_linked":[{"id":"e5ce53c8-04f2-46d0-81a3-d2e814dc1c91","orcid":null,"display_name":"Bowen Jin"},{"id":"cdf22883-7343-4c95-960b-5ac6692233de","orcid":null,"display_name":"Hansi Zeng"},{"id":"8a201a69-5011-4650-95db-d5eb0e8a2a35","orcid":null,"display_name":"Zhenrui Yue"},{"id":"8eabd1c5-5cff-4f3c-956c-a74b618d0e83","orcid":null,"display_name":"Jinsung Yoon"},{"id":"16281cbd-a22f-4a34-95a1-76a2b363672e","orcid":null,"display_name":"Sercan Arik"},{"id":"cefe005c-2766-45d4-8599-fc8e99930836","orcid":null,"display_name":"Dong Wang"}]},"error":null,"updated_at":"2026-05-18T09:30:40.308968+00:00"},"context_extract":{"job_type":"context_extract","status":"succeeded","result":{"enqueued_papers":25},"error":null,"updated_at":"2026-05-14T08:07:52.214680+00:00"},"graph_features":{"job_type":"graph_features","status":"succeeded","result":{"co_cited":[{"title":"DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models","work_id":"c5006563-f3ec-438a-9e35-b7b484f34828","shared_citers":41},{"title":"Qwen3 Technical Report","work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e","shared_citers":32},{"title":"DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning","work_id":"e6b75ad5-2877-4168-97c8-710407094d20","shared_citers":29},{"title":"Proximal Policy Optimization Algorithms","work_id":"240c67fe-d14d-4520-91c1-38a4e272ca19","shared_citers":28},{"title":"DAPO: An Open-Source LLM Reinforcement Learning System at Scale","work_id":"64019d00-0b11-4bbd-b173-b46c8fad0157","shared_citers":19},{"title":"Qwen2.5 Technical Report","work_id":"d8432992-4980-4a81-85c7-9fa2c2b87f85","shared_citers":18},{"title":"Search-o1: Agentic Search-Enhanced Large Reasoning Models","work_id":"6246e99b-7b4e-4424-be78-3e1983932cdc","shared_citers":16},{"title":"R1-Searcher: Incentivizing the Search Capability in LLMs via Reinforcement Learning","work_id":"f5ed73d2-f2ff-4cf6-853d-3586333e44ef","shared_citers":13},{"title":"Text Embeddings by Weakly-Supervised Contrastive Pre-training","work_id":"789cc674-467e-4f23-bb50-05c79fe8c4c2","shared_citers":13},{"title":"OpenAI o1 System Card","work_id":"68d3c334-0fc9-49e3-b7b0-a69afae933e2","shared_citers":12},{"title":"RAGEN: Understanding Self-Evolution in LLM Agents via Multi-Turn Reinforcement Learning","work_id":"b96383ee-f8dc-471f-aba4-bc5ce9b0b632","shared_citers":11},{"title":"The Llama 3 Herd of Models","work_id":"1549a635-88af-4ac1-acfe-51ae7bb53345","shared_citers":11},{"title":"DeepSeek-V3 Technical Report","work_id":"57d2791d-2219-4c31-a077-afc04b12a75c","shared_citers":10},{"title":"GPT-4o System Card","work_id":"f37bf1c7-4964-4e56-9762-d20da8d9009f","shared_citers":10},{"title":"Kimi k1.5: Scaling Reinforcement Learning with LLMs","work_id":"bff96ab1-bd6a-4585-be23-74fdb51969c7","shared_citers":10},{"title":"Training Verifiers to Solve Math Word Problems","work_id":"acab1aa8-b4d6-40e0-a3ee-25341701dca2","shared_citers":10},{"title":"Group-in-Group Policy Optimization for LLM Agent Training","work_id":"bc65d492-e6ba-4522-874c-43d2f4fc5191","shared_citers":9},{"title":"Qwen2.5-VL Technical Report","work_id":"69dffacb-bfe8-442d-be86-48624c60426f","shared_citers":9},{"title":"Understanding R1-Zero-Like Training: A Critical Perspective","work_id":"ec354f3b-9484-4a0c-94c8-92d4d0260835","shared_citers":9},{"title":"arXiv preprint arXiv:2504.03160 , year=","work_id":"460d8021-a193-45c3-89a7-b72f6767c87f","shared_citers":8},{"title":"From Local to Global: A Graph RAG Approach to Query-Focused Summarization","work_id":"588618d7-fd41-4053-b34d-a981f8793039","shared_citers":7},{"title":"Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities","work_id":"008df105-2fdd-45d8-857a-8e35868aecb6","shared_citers":7},{"title":"GPT-4 Technical Report","work_id":"b928e041-6991-4c08-8c81-0359e4097c7b","shared_citers":7},{"title":"Group Sequence Policy Optimization","work_id":"3a98b53b-9f52-4d95-adf7-89353c0a9a65","shared_citers":7}],"time_series":[{"n":3,"year":2025},{"n":73,"year":2026}],"dependency_candidates":[]},"error":null,"updated_at":"2026-05-14T08:07:46.017672+00:00"},"identity_refresh":{"job_type":"identity_refresh","status":"succeeded","result":{"items":[{"title":"Qwen3 Technical Report","outcome":"unchanged","work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e","resolver":"local_arxiv","confidence":0.98,"old_work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e"}],"counts":{"fixed":0,"merged":0,"unchanged":1,"quarantined":0,"needs_external_resolution":0},"errors":[],"attempted":1},"error":null,"updated_at":"2026-05-14T08:07:43.608981+00:00"},"role_polarity":{"job_type":"role_polarity","status":"succeeded","result":{"title":"Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning","claims":[{"claim_text":"Efficiently acquiring external knowledge and up-to-date information is essential for effective reasoning and text generation in large language models (LLMs). Prompting advanced LLMs with reasoning capabilities to use search engines during inference is often suboptimal, as the LLM might not fully possess the capability on how to interact optimally with the search engine. This paper introduces Search-R1, an extension of reinforcement learning (RL) for reasoning frameworks where the LLM learns to autonomously generate (multiple) search queries during step-by-step reasoning with real-time retrieva","claim_type":"abstract","evidence_strength":"source_metadata"},{"claim_text":"turn search horizons, achieving state-of-the-art performance among open-source models on challenging long-horizon benchmarks like MM-BrowseComp and MMSearch-Plus, while also exhibiting strong generalizability across different base models. Our code will be released in https://github.com/RUCAIBox/ LMM-Searcher. 1 Introduction Deep search agent systems [1, 2, 3] have achieved significant success in tackling challenging real- world information-seeking problems [4, 5]. Building on the deep search fra","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"org/abs/2402.01680. [9] Jie Huang and Kevin Chen-Chuan Chang. Towards reasoning in large language models: A survey. InFindings of the Association for Computational Linguistics: ACL 2023, pages 1049-1065, Toronto, Canada, 2023. Association for Computational Linguistics. doi: 10.18653/v1/2023.findings-acl.67. URL https://aclanthology.org/2023.findings-acl.67/. [10] Bowen Jin, Hansi Zeng, Zhenrui Yue, Jinsung Yoon, Sercan Arik, Dong Wang, Hamed Zamani, and Jiawei Han. Search-r1: Training llms to re","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"We evaluate three representative RL pipelines for agent training.Single-Turngenerates a complete response in one step, commonly used for reasoning tasks [16, 21]. Multi-Turniteratively produces intermediate outputs for self- refinement and multi-agent interaction [8, 25, 41, 53].Tool- Usageinvokes external tools during rollout; we use the Search-R1 pipeline [23]. We adopt GRPO [43] for policy optimization, with a global batch size of 64, group size 4, and maximum response length of 8192 tokens. ","claim_type":"method","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"Xu, Zhongyu Zhang, and Zhen Zhang. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning, 2025. URL https://arxiv.org/abs/2501.12948. [5] Luyu Gao, Aman Madaan, Shuyan Zhou, Uri Alon, Pengfei Liu, Yiming Yang, Jamie Callan, and Graham Neubig. Pal: Program-aided language models, 2023. URLhttps://arxiv.org/abs/2211.10435. [6] Bowen Jin, Hansi Zeng, Zhenrui Yue, Jinsung Yoon, Sercan Arik, Dong Wang, Hamed Zamani, and Jiawei Han. Search-r1: Training llms to reason and le","claim_type":"other","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"RL algorithms have emerged as promising alternatives to traditional methods like PPO [42]. These methods, such as RLOO [16, 17], GRPO [18], Dr. GRPO [19], DAPO [20], and CPPO [57], avoid introducing extra value functions by leveraging a group of samples from the same query and estimate the advantages accordingly. This enables the large-scale RL training and has shown strong results in tasks such as mathematical reasoning [15], search [58, 59], and tool use [60, 61]. Our work is closely related t","claim_type":"baseline","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"ods commonly adopt a translation-based strategy [1, 2, 13, 28, 31], translating documents into a single language before retrieval and generation. This approach reduces language gaps to some extent, but noise introduced by translation errors may lead to the loss of key information. Other methods focus on modifying the knowl- edge injection process [8, 41] or the reasoning strategy [16, 19, 32]. However, modifying knowledge injection often disrupts semantic integrity, while complex reasoning strat","claim_type":"background","confidence":0.85,"evidence_strength":"citation_context"}],"why_cited":"Pith tracks Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning because it crossed a citation-hub threshold. Current citing contexts most often use it as background evidence (23 contexts).","role_counts":[{"n":23,"context_role":"background"},{"n":3,"context_role":"method"},{"n":1,"context_role":"baseline"},{"n":1,"context_role":"other"}]},"error":null,"updated_at":"2026-05-18T09:30:40.311996+00:00"},"summary_claims":{"job_type":"summary_claims","status":"succeeded","result":{"title":"Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning","claims":[{"claim_text":"Efficiently acquiring external knowledge and up-to-date information is essential for effective reasoning and text generation in large language models (LLMs). Prompting advanced LLMs with reasoning capabilities to use search engines during inference is often suboptimal, as the LLM might not fully possess the capability on how to interact optimally with the search engine. This paper introduces Search-R1, an extension of reinforcement learning (RL) for reasoning frameworks where the LLM learns to autonomously generate (multiple) search queries during step-by-step reasoning with real-time retrieva","claim_type":"abstract","evidence_strength":"source_metadata"}],"why_cited":"Pith tracks Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning because it crossed a citation-hub threshold.","role_counts":[]},"error":null,"updated_at":"2026-05-14T08:07:41.235650+00:00"}},"summary":{"title":"Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning","claims":[{"claim_text":"Efficiently acquiring external knowledge and up-to-date information is essential for effective reasoning and text generation in large language models (LLMs). Prompting advanced LLMs with reasoning capabilities to use search engines during inference is often suboptimal, as the LLM might not fully possess the capability on how to interact optimally with the search engine. This paper introduces Search-R1, an extension of reinforcement learning (RL) for reasoning frameworks where the LLM learns to autonomously generate (multiple) search queries during step-by-step reasoning with real-time retrieva","claim_type":"abstract","evidence_strength":"source_metadata"}],"why_cited":"Pith tracks Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning because it crossed a citation-hub threshold.","role_counts":[]},"graph":{"co_cited":[{"title":"DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models","work_id":"c5006563-f3ec-438a-9e35-b7b484f34828","shared_citers":41},{"title":"Qwen3 Technical Report","work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e","shared_citers":32},{"title":"DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning","work_id":"e6b75ad5-2877-4168-97c8-710407094d20","shared_citers":29},{"title":"Proximal Policy Optimization Algorithms","work_id":"240c67fe-d14d-4520-91c1-38a4e272ca19","shared_citers":28},{"title":"DAPO: An Open-Source LLM Reinforcement Learning System at Scale","work_id":"64019d00-0b11-4bbd-b173-b46c8fad0157","shared_citers":19},{"title":"Qwen2.5 Technical Report","work_id":"d8432992-4980-4a81-85c7-9fa2c2b87f85","shared_citers":18},{"title":"Search-o1: Agentic Search-Enhanced Large Reasoning Models","work_id":"6246e99b-7b4e-4424-be78-3e1983932cdc","shared_citers":16},{"title":"R1-Searcher: Incentivizing the Search Capability in LLMs via Reinforcement Learning","work_id":"f5ed73d2-f2ff-4cf6-853d-3586333e44ef","shared_citers":13},{"title":"Text Embeddings by Weakly-Supervised Contrastive Pre-training","work_id":"789cc674-467e-4f23-bb50-05c79fe8c4c2","shared_citers":13},{"title":"OpenAI o1 System Card","work_id":"68d3c334-0fc9-49e3-b7b0-a69afae933e2","shared_citers":12},{"title":"RAGEN: Understanding Self-Evolution in LLM Agents via Multi-Turn Reinforcement Learning","work_id":"b96383ee-f8dc-471f-aba4-bc5ce9b0b632","shared_citers":11},{"title":"The Llama 3 Herd of Models","work_id":"1549a635-88af-4ac1-acfe-51ae7bb53345","shared_citers":11},{"title":"DeepSeek-V3 Technical Report","work_id":"57d2791d-2219-4c31-a077-afc04b12a75c","shared_citers":10},{"title":"GPT-4o System Card","work_id":"f37bf1c7-4964-4e56-9762-d20da8d9009f","shared_citers":10},{"title":"Kimi k1.5: Scaling Reinforcement Learning with LLMs","work_id":"bff96ab1-bd6a-4585-be23-74fdb51969c7","shared_citers":10},{"title":"Training Verifiers to Solve Math Word Problems","work_id":"acab1aa8-b4d6-40e0-a3ee-25341701dca2","shared_citers":10},{"title":"Group-in-Group Policy Optimization for LLM Agent Training","work_id":"bc65d492-e6ba-4522-874c-43d2f4fc5191","shared_citers":9},{"title":"Qwen2.5-VL Technical Report","work_id":"69dffacb-bfe8-442d-be86-48624c60426f","shared_citers":9},{"title":"Understanding R1-Zero-Like Training: A Critical Perspective","work_id":"ec354f3b-9484-4a0c-94c8-92d4d0260835","shared_citers":9},{"title":"arXiv preprint arXiv:2504.03160 , year=","work_id":"460d8021-a193-45c3-89a7-b72f6767c87f","shared_citers":8},{"title":"From Local to Global: A Graph RAG Approach to Query-Focused Summarization","work_id":"588618d7-fd41-4053-b34d-a981f8793039","shared_citers":7},{"title":"Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities","work_id":"008df105-2fdd-45d8-857a-8e35868aecb6","shared_citers":7},{"title":"GPT-4 Technical Report","work_id":"b928e041-6991-4c08-8c81-0359e4097c7b","shared_citers":7},{"title":"Group Sequence Policy Optimization","work_id":"3a98b53b-9f52-4d95-adf7-89353c0a9a65","shared_citers":7}],"time_series":[{"n":3,"year":2025},{"n":73,"year":2026}],"dependency_candidates":[]},"authors":[{"id":"e5ce53c8-04f2-46d0-81a3-d2e814dc1c91","orcid":null,"display_name":"Bowen Jin","source":"manual","import_confidence":0.72},{"id":"cefe005c-2766-45d4-8599-fc8e99930836","orcid":null,"display_name":"Dong Wang","source":"manual","import_confidence":0.72},{"id":"cdf22883-7343-4c95-960b-5ac6692233de","orcid":null,"display_name":"Hansi Zeng","source":"manual","import_confidence":0.72},{"id":"8eabd1c5-5cff-4f3c-956c-a74b618d0e83","orcid":null,"display_name":"Jinsung Yoon","source":"manual","import_confidence":0.72},{"id":"16281cbd-a22f-4a34-95a1-76a2b363672e","orcid":null,"display_name":"Sercan Arik","source":"manual","import_confidence":0.72},{"id":"8a201a69-5011-4650-95db-d5eb0e8a2a35","orcid":null,"display_name":"Zhenrui Yue","source":"manual","import_confidence":0.72}]}}