{"work":{"id":"feef9556-a016-493c-abd2-0c97a23a7ebf","openalex_id":null,"doi":null,"arxiv_id":"2404.14219","raw_key":null,"title":"Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone","authors":null,"authors_text":"Marah Abdin, Jyoti Aneja, Hany Awadalla, Ahmed Awadallah, Ammar Ahmad Awan, Nguyen Bach","year":2024,"venue":"cs.CL","abstract":"We introduce phi-3-mini, a 3.8 billion parameter language model trained on 3.3 trillion tokens, whose overall performance, as measured by both academic benchmarks and internal testing, rivals that of models such as Mixtral 8x7B and GPT-3.5 (e.g., phi-3-mini achieves 69% on MMLU and 8.38 on MT-bench), despite being small enough to be deployed on a phone. Our training dataset is a scaled-up version of the one used for phi-2, composed of heavily filtered publicly available web data and synthetic data. The model is also further aligned for robustness, safety, and chat format. We also provide parameter-scaling results with a 7B, 14B models trained for 4.8T tokens, called phi-3-small, phi-3-medium, both significantly more capable than phi-3-mini (e.g., respectively 75%, 78% on MMLU, and 8.7, 8.9 on MT-bench). To enhance multilingual, multimodal, and long-context capabilities, we introduce three models in the phi-3.5 series: phi-3.5-mini, phi-3.5-MoE, and phi-3.5-Vision. The phi-3.5-MoE, a 16 x 3.8B MoE model with 6.6 billion active parameters, achieves superior performance in language reasoning, math, and code tasks compared to other open-source models of similar scale, such as Llama 3.1 and the Mixtral series, and on par with Gemini-1.5-Flash and GPT-4o-mini. Meanwhile, phi-3.5-Vision, a 4.2 billion parameter model derived from phi-3.5-mini, excels in reasoning tasks and is adept at handling both single-image and text prompts, as well as multi-image and text prompts.","external_url":"https://arxiv.org/abs/2404.14219","cited_by_count":null,"metadata_source":"pith","metadata_fetched_at":"2026-05-25T06:15:23.695197+00:00","pith_arxiv_id":"2404.14219","created_at":"2026-05-09T06:05:35.478025+00:00","updated_at":"2026-05-25T06:15:23.695197+00:00","title_quality_ok":true,"display_title":"Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone","render_title":"Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone"},"hub":{"state":{"work_id":"feef9556-a016-493c-abd2-0c97a23a7ebf","tier":"super_hub","tier_reason":"100+ Pith inbound or 10,000+ external citations","pith_inbound_count":149,"external_cited_by_count":null,"distinct_field_count":17,"first_pith_cited_at":"2023-03-31T17:28:46+00:00","last_pith_cited_at":"2026-05-22T17:16:35+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":"2026-06-04T03:06:55.138420+00:00","tier_text":"super_hub"},"tier":"super_hub","role_counts":[{"context_role":"background","n":25},{"context_role":"baseline","n":9},{"context_role":"method","n":4},{"context_role":"dataset","n":2}],"polarity_counts":[{"context_polarity":"background","n":25},{"context_polarity":"baseline","n":9},{"context_polarity":"use_method","n":4},{"context_polarity":"use_dataset","n":2}],"runs":{"ask_index":{"job_type":"ask_index","status":"succeeded","result":{"title":"Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone","claims":[{"claim_text":"We introduce phi-3-mini, a 3.8 billion parameter language model trained on 3.3 trillion tokens, whose overall performance, as measured by both academic benchmarks and internal testing, rivals that of models such as Mixtral 8x7B and GPT-3.5 (e.g., phi-3-mini achieves 69% on MMLU and 8.38 on MT-bench), despite being small enough to be deployed on a phone. Our training dataset is a scaled-up version of the one used for phi-2, composed of heavily filtered publicly available web data and synthetic data. The model is also further aligned for robustness, safety, and chat format. We also provide param","claim_type":"abstract","evidence_strength":"source_metadata"},{"claim_text":"We integrate two rounds of expert reviews to reduce the noise of the dataset. The first round is based on expert verification. In the second round, we utilize the SoTA LLMs to identify potential errors and employ annotators to perform more targeted verification. We evaluated more than 50 LLMs including open-source and closed-source models, such as GPT- 4o [17], Claude-3-Opus [6], and Gemini [30], LLaMA-3 [35], Phi-3 [1] on MMLU-Pro. Our key findings are summarized as follows: 1. MMLU-Pro present","claim_type":"baseline","confidence":0.95,"evidence_strength":"citation_context"},{"claim_text":"For the datasets introduced in the previous section, we generated about 5M samples for each modality, except for the GUI modality, for which we generated 30M samples. Since our data are primarily intended for the pre-training or mid-training stages of VLMs, we require a base model that has not been exposed to GUI-related pre-training. To this end, we adopt Phi-3.5-VL [43], a 4B-parameter VLM, as the backbone. We put the detailed hyper-parameters and data proportion in Appendix B. 5.2 Benchmarks ","claim_type":"method","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"Due to space limits, we present detailed descriptions of experimental settings in Appendix C. Table 1Evaluation results on RULER [28]. Average accuracy (%) is reported, with the best results inbold. In-Domain Evaluation Extrapolation Model 4k 8k 16k 32k 64k 128k 256k Mistral-7B [31]93.6 91.2 87.2 75.4 49.0 13.8 - GLM3-6B [23]87.8 83.4 78.6 69.9 56.0 42.0 - Phi3-medium-14B [1]93.3 93.2 91.1 86.8 78.6 46.1 - Llama3-8B [41]92.8 90.3 85.7 79.9 76.3 69.5 - Qwen3-4B (Instruct) [57]95.1 93.6 91.0 87.8 ","claim_type":"baseline","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"cloud-based or server-side inference pipelines in which patient conversational data is transmitted to remote compute infrastructure for processing [9, 7]. This architectural dependency creates an irreconcilable tension with the pri- vacy requirements of the most at-risk and underserved populations. The emergence of lightweight, highly efficient open-source LLMs-such as Gemma [10], Phi-3.5-mini [11], and Qwen2 [12]-combined with advances in model compression via Quantized Low-Rank Adaptation (QLo","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"• Our model achieves significantly better performance than previous VLAs, exhibiting quick adaptation to new robots and tasks and effective generation to un- seen objects and backgrounds, as shown in Figure 1. All our code and models are publicly released. 2. Related Works Vision-Language-Action Models. The success of Large Language Models (LLMs) [2, 9, 63, 64] and Vision- Language Models (VLMs) [1, 14, 28, 37, 61] has inspired the development of Vision-Language-Action (VLA) mod- els, which exte","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"we ultimately demonstrate that models trained on FineWeb perform better than those trained on other 38th Conference on Neural Information Processing Systems (NeurIPS 2024) Track on Datasets and Benchmarks. arXiv:2406.17557v2 [cs.CL] 31 Oct 2024 public web-based pre-training datasets. Inspired by recent work advocating for training LLMs on educational data [8, 9], we additionally introduce FineWeb-Edu, a subset of 1.3 trillion tokens from FineWeb that was rated as highly educational by a custom c","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"}],"why_cited":"Pith tracks Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone because it crossed a citation-hub threshold. Current citing contexts most often use it as background evidence (11 contexts).","role_counts":[{"n":11,"context_role":"background"},{"n":6,"context_role":"baseline"},{"n":3,"context_role":"method"},{"n":1,"context_role":"dataset"}]},"error":null,"updated_at":"2026-05-17T23:00:23.460309+00:00"},"author_expand":{"job_type":"author_expand","status":"succeeded","result":{"authors_linked":[{"id":"2df2d991-a259-4c92-9e30-f428e986de0c","orcid":null,"display_name":"Marah Abdin"},{"id":"af8698dc-c88d-447f-91a0-e07de89bdd20","orcid":null,"display_name":"Jyoti Aneja"},{"id":"d32226f7-aabb-41fe-b148-cf602e8b67ae","orcid":null,"display_name":"Hany Awadalla"},{"id":"a098f08b-5567-43c0-9dd3-b5546218cee2","orcid":null,"display_name":"Ahmed Awadallah"},{"id":"5a2fc288-3a69-48a4-bf85-5ed180028773","orcid":null,"display_name":"Ammar Ahmad Awan"},{"id":"05afd7b5-d16c-4050-ba3a-8328cb7f9135","orcid":null,"display_name":"Nguyen Bach"}]},"error":null,"updated_at":"2026-05-17T23:00:23.449784+00:00"},"context_extract":{"job_type":"context_extract","status":"succeeded","result":{"enqueued_papers":25},"error":null,"updated_at":"2026-05-14T09:58:43.874758+00:00"},"graph_features":{"job_type":"graph_features","status":"succeeded","result":{"co_cited":[{"title":"The Llama 3 Herd of Models","work_id":"1549a635-88af-4ac1-acfe-51ae7bb53345","shared_citers":22},{"title":"GPT-4 Technical Report","work_id":"b928e041-6991-4c08-8c81-0359e4097c7b","shared_citers":14},{"title":"Qwen2.5 Technical Report","work_id":"d8432992-4980-4a81-85c7-9fa2c2b87f85","shared_citers":14},{"title":"Qwen3 Technical Report","work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e","shared_citers":12},{"title":"Mistral 7B","work_id":"eb5e1305-ad11-4875-ad8d-ad8b8f697599","shared_citers":11},{"title":"Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context","work_id":"80e3e977-f1bb-4c83-8d0c-1ab0a0c5c3f1","shared_citers":10},{"title":"Llama 2: Open Foundation and Fine-Tuned Chat Models","work_id":"68a5177f-d644-44c1-bd4f-4e5278c22f5d","shared_citers":10},{"title":"Evaluating Large Language Models Trained on Code","work_id":"042493e9-b26f-4b4e-bbde-382072ca9b08","shared_citers":9},{"title":"Mixtral of Experts","work_id":"0de8c352-9daa-4e1e-8c7b-3d0dec69f369","shared_citers":9},{"title":"Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution","work_id":"8abcfe4f-e0fb-44b7-9123-448fac95f90a","shared_citers":9},{"title":"Scaling Laws for Neural Language Models","work_id":"b7dd8749-9c45-4977-ab9b-64478dce1ae8","shared_citers":9},{"title":"Gemini: A Family of Highly Capable Multimodal Models","work_id":"83f7c85b-3f11-450f-ac0c-64d9745220b2","shared_citers":8},{"title":"Gemma 3 Technical Report","work_id":"f93e08bf-9e96-409b-8ac6-b8385fd17fd7","shared_citers":8},{"title":"Gemma: Open Models Based on Gemini Research and Technology","work_id":"a9ea2870-df28-40b8-a9e0-a7e9a116f793","shared_citers":8},{"title":"GPT-4o System Card","work_id":"f37bf1c7-4964-4e56-9762-d20da8d9009f","shared_citers":8},{"title":"DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning","work_id":"e6b75ad5-2877-4168-97c8-710407094d20","shared_citers":7},{"title":"Gemma 2: Improving Open Language Models at a Practical Size","work_id":"4dd94e2f-2b27-4cbf-88a0-4910f0772a57","shared_citers":7},{"title":"Language Models are Few-Shot Learners","work_id":"214732c0-2edd-44a0-af9e-28184a2b8279","shared_citers":7},{"title":"LLaMA: Open and Efficient Foundation Language Models","work_id":"c018fc23-6f3f-4035-9d02-28a2173b2b9d","shared_citers":7},{"title":"LoRA: Low-Rank Adaptation of Large Language Models","work_id":"0426219a-789e-4964-adc8-a04538510818","shared_citers":7},{"title":"Program Synthesis with Large Language Models","work_id":"fd241a05-03b9-4de2-9588-9d77ce176125","shared_citers":7},{"title":"Qwen2 Technical Report","work_id":"a1857881-ab9b-4b80-9b5f-9ae4b5c2566d","shared_citers":7},{"title":"Training Verifiers to Solve Math Word Problems","work_id":"acab1aa8-b4d6-40e0-a3ee-25341701dca2","shared_citers":7},{"title":"Yi: Open Foundation Models by 01.AI","work_id":"8efee8a1-5e3c-4851-9c65-18e3d1d9e769","shared_citers":7}],"time_series":[{"n":1,"year":2023},{"n":11,"year":2024},{"n":4,"year":2025},{"n":48,"year":2026}],"dependency_candidates":[]},"error":null,"updated_at":"2026-05-14T09:58:47.743983+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":"2026-05-14T09:58:39.606017+00:00"},"role_polarity":{"job_type":"role_polarity","status":"succeeded","result":{"title":"Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone","claims":[{"claim_text":"We introduce phi-3-mini, a 3.8 billion parameter language model trained on 3.3 trillion tokens, whose overall performance, as measured by both academic benchmarks and internal testing, rivals that of models such as Mixtral 8x7B and GPT-3.5 (e.g., phi-3-mini achieves 69% on MMLU and 8.38 on MT-bench), despite being small enough to be deployed on a phone. Our training dataset is a scaled-up version of the one used for phi-2, composed of heavily filtered publicly available web data and synthetic data. The model is also further aligned for robustness, safety, and chat format. We also provide param","claim_type":"abstract","evidence_strength":"source_metadata"},{"claim_text":"We integrate two rounds of expert reviews to reduce the noise of the dataset. The first round is based on expert verification. In the second round, we utilize the SoTA LLMs to identify potential errors and employ annotators to perform more targeted verification. We evaluated more than 50 LLMs including open-source and closed-source models, such as GPT- 4o [17], Claude-3-Opus [6], and Gemini [30], LLaMA-3 [35], Phi-3 [1] on MMLU-Pro. Our key findings are summarized as follows: 1. MMLU-Pro present","claim_type":"baseline","confidence":0.95,"evidence_strength":"citation_context"},{"claim_text":"For the datasets introduced in the previous section, we generated about 5M samples for each modality, except for the GUI modality, for which we generated 30M samples. Since our data are primarily intended for the pre-training or mid-training stages of VLMs, we require a base model that has not been exposed to GUI-related pre-training. To this end, we adopt Phi-3.5-VL [43], a 4B-parameter VLM, as the backbone. We put the detailed hyper-parameters and data proportion in Appendix B. 5.2 Benchmarks ","claim_type":"method","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"Due to space limits, we present detailed descriptions of experimental settings in Appendix C. Table 1Evaluation results on RULER [28]. Average accuracy (%) is reported, with the best results inbold. In-Domain Evaluation Extrapolation Model 4k 8k 16k 32k 64k 128k 256k Mistral-7B [31]93.6 91.2 87.2 75.4 49.0 13.8 - GLM3-6B [23]87.8 83.4 78.6 69.9 56.0 42.0 - Phi3-medium-14B [1]93.3 93.2 91.1 86.8 78.6 46.1 - Llama3-8B [41]92.8 90.3 85.7 79.9 76.3 69.5 - Qwen3-4B (Instruct) [57]95.1 93.6 91.0 87.8 ","claim_type":"baseline","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"cloud-based or server-side inference pipelines in which patient conversational data is transmitted to remote compute infrastructure for processing [9, 7]. This architectural dependency creates an irreconcilable tension with the pri- vacy requirements of the most at-risk and underserved populations. The emergence of lightweight, highly efficient open-source LLMs-such as Gemma [10], Phi-3.5-mini [11], and Qwen2 [12]-combined with advances in model compression via Quantized Low-Rank Adaptation (QLo","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"• Our model achieves significantly better performance than previous VLAs, exhibiting quick adaptation to new robots and tasks and effective generation to un- seen objects and backgrounds, as shown in Figure 1. All our code and models are publicly released. 2. Related Works Vision-Language-Action Models. The success of Large Language Models (LLMs) [2, 9, 63, 64] and Vision- Language Models (VLMs) [1, 14, 28, 37, 61] has inspired the development of Vision-Language-Action (VLA) mod- els, which exte","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"},{"claim_text":"we ultimately demonstrate that models trained on FineWeb perform better than those trained on other 38th Conference on Neural Information Processing Systems (NeurIPS 2024) Track on Datasets and Benchmarks. arXiv:2406.17557v2 [cs.CL] 31 Oct 2024 public web-based pre-training datasets. Inspired by recent work advocating for training LLMs on educational data [8, 9], we additionally introduce FineWeb-Edu, a subset of 1.3 trillion tokens from FineWeb that was rated as highly educational by a custom c","claim_type":"background","confidence":0.9,"evidence_strength":"citation_context"}],"why_cited":"Pith tracks Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone because it crossed a citation-hub threshold. Current citing contexts most often use it as background evidence (11 contexts).","role_counts":[{"n":11,"context_role":"background"},{"n":6,"context_role":"baseline"},{"n":3,"context_role":"method"},{"n":1,"context_role":"dataset"}]},"error":null,"updated_at":"2026-05-17T23:00:23.456096+00:00"},"summary_claims":{"job_type":"summary_claims","status":"succeeded","result":{"title":"Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone","claims":[{"claim_text":"We introduce phi-3-mini, a 3.8 billion parameter language model trained on 3.3 trillion tokens, whose overall performance, as measured by both academic benchmarks and internal testing, rivals that of models such as Mixtral 8x7B and GPT-3.5 (e.g., phi-3-mini achieves 69% on MMLU and 8.38 on MT-bench), despite being small enough to be deployed on a phone. Our training dataset is a scaled-up version of the one used for phi-2, composed of heavily filtered publicly available web data and synthetic data. The model is also further aligned for robustness, safety, and chat format. We also provide param","claim_type":"abstract","evidence_strength":"source_metadata"}],"why_cited":"Pith tracks Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone because it crossed a citation-hub threshold.","role_counts":[]},"error":null,"updated_at":"2026-05-14T09:58:41.951094+00:00"}},"summary":{"title":"Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone","claims":[{"claim_text":"We introduce phi-3-mini, a 3.8 billion parameter language model trained on 3.3 trillion tokens, whose overall performance, as measured by both academic benchmarks and internal testing, rivals that of models such as Mixtral 8x7B and GPT-3.5 (e.g., phi-3-mini achieves 69% on MMLU and 8.38 on MT-bench), despite being small enough to be deployed on a phone. Our training dataset is a scaled-up version of the one used for phi-2, composed of heavily filtered publicly available web data and synthetic data. The model is also further aligned for robustness, safety, and chat format. We also provide param","claim_type":"abstract","evidence_strength":"source_metadata"}],"why_cited":"Pith tracks Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone because it crossed a citation-hub threshold.","role_counts":[]},"graph":{"co_cited":[{"title":"The Llama 3 Herd of Models","work_id":"1549a635-88af-4ac1-acfe-51ae7bb53345","shared_citers":22},{"title":"GPT-4 Technical Report","work_id":"b928e041-6991-4c08-8c81-0359e4097c7b","shared_citers":14},{"title":"Qwen2.5 Technical Report","work_id":"d8432992-4980-4a81-85c7-9fa2c2b87f85","shared_citers":14},{"title":"Qwen3 Technical Report","work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e","shared_citers":12},{"title":"Mistral 7B","work_id":"eb5e1305-ad11-4875-ad8d-ad8b8f697599","shared_citers":11},{"title":"Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context","work_id":"80e3e977-f1bb-4c83-8d0c-1ab0a0c5c3f1","shared_citers":10},{"title":"Llama 2: Open Foundation and Fine-Tuned Chat Models","work_id":"68a5177f-d644-44c1-bd4f-4e5278c22f5d","shared_citers":10},{"title":"Evaluating Large Language Models Trained on Code","work_id":"042493e9-b26f-4b4e-bbde-382072ca9b08","shared_citers":9},{"title":"Mixtral of Experts","work_id":"0de8c352-9daa-4e1e-8c7b-3d0dec69f369","shared_citers":9},{"title":"Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution","work_id":"8abcfe4f-e0fb-44b7-9123-448fac95f90a","shared_citers":9},{"title":"Scaling Laws for Neural Language Models","work_id":"b7dd8749-9c45-4977-ab9b-64478dce1ae8","shared_citers":9},{"title":"Gemini: A Family of Highly Capable Multimodal Models","work_id":"83f7c85b-3f11-450f-ac0c-64d9745220b2","shared_citers":8},{"title":"Gemma 3 Technical Report","work_id":"f93e08bf-9e96-409b-8ac6-b8385fd17fd7","shared_citers":8},{"title":"Gemma: Open Models Based on Gemini Research and Technology","work_id":"a9ea2870-df28-40b8-a9e0-a7e9a116f793","shared_citers":8},{"title":"GPT-4o System Card","work_id":"f37bf1c7-4964-4e56-9762-d20da8d9009f","shared_citers":8},{"title":"DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning","work_id":"e6b75ad5-2877-4168-97c8-710407094d20","shared_citers":7},{"title":"Gemma 2: Improving Open Language Models at a Practical Size","work_id":"4dd94e2f-2b27-4cbf-88a0-4910f0772a57","shared_citers":7},{"title":"Language Models are Few-Shot Learners","work_id":"214732c0-2edd-44a0-af9e-28184a2b8279","shared_citers":7},{"title":"LLaMA: Open and Efficient Foundation Language Models","work_id":"c018fc23-6f3f-4035-9d02-28a2173b2b9d","shared_citers":7},{"title":"LoRA: Low-Rank Adaptation of Large Language Models","work_id":"0426219a-789e-4964-adc8-a04538510818","shared_citers":7},{"title":"Program Synthesis with Large Language Models","work_id":"fd241a05-03b9-4de2-9588-9d77ce176125","shared_citers":7},{"title":"Qwen2 Technical Report","work_id":"a1857881-ab9b-4b80-9b5f-9ae4b5c2566d","shared_citers":7},{"title":"Training Verifiers to Solve Math Word Problems","work_id":"acab1aa8-b4d6-40e0-a3ee-25341701dca2","shared_citers":7},{"title":"Yi: Open Foundation Models by 01.AI","work_id":"8efee8a1-5e3c-4851-9c65-18e3d1d9e769","shared_citers":7}],"time_series":[{"n":1,"year":2023},{"n":11,"year":2024},{"n":4,"year":2025},{"n":48,"year":2026}],"dependency_candidates":[]},"authors":[{"id":"a098f08b-5567-43c0-9dd3-b5546218cee2","orcid":null,"display_name":"Ahmed Awadallah","source":"manual","import_confidence":0.72},{"id":"5a2fc288-3a69-48a4-bf85-5ed180028773","orcid":null,"display_name":"Ammar Ahmad Awan","source":"manual","import_confidence":0.72},{"id":"d32226f7-aabb-41fe-b148-cf602e8b67ae","orcid":null,"display_name":"Hany Awadalla","source":"manual","import_confidence":0.72},{"id":"af8698dc-c88d-447f-91a0-e07de89bdd20","orcid":null,"display_name":"Jyoti Aneja","source":"manual","import_confidence":0.72},{"id":"2df2d991-a259-4c92-9e30-f428e986de0c","orcid":null,"display_name":"Marah Abdin","source":"manual","import_confidence":0.72},{"id":"05afd7b5-d16c-4050-ba3a-8328cb7f9135","orcid":null,"display_name":"Nguyen Bach","source":"manual","import_confidence":0.72}]}}