[{"data":1,"prerenderedAt":176},["ShallowReactive",2],{"content-query-SHGd64GqJl":3},{"_path":4,"_dir":5,"_draft":6,"_partial":6,"_locale":7,"title":8,"description":9,"date":13,"cover":14,"type":15,"body":16,"_type":170,"_id":171,"_source":172,"_file":173,"_stem":174,"_extension":175},"/news/en/2828","en",false,"","Baichuan AI x Ascend AI | Baichuan 2 Open Sourced in the MindSpore Community",[10,11,12],"Beijing","September 6",2023,"2023-09-06","https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2023/10/24/c874fc091c7142d3a682876d1e87a2d7.png","news",{"type":17,"children":18,"toc":167},"root",[19,27,33,38,46,51,56,67,72,81,90,95,102,107,112,119,124,131,136,141,149,154,162],{"type":20,"tag":21,"props":22,"children":24},"element","h1",{"id":23},"baichuan-ai-x-ascend-ai-baichuan-2-open-sourced-in-the-mindspore-community",[25],{"type":26,"value":8},"text",{"type":20,"tag":28,"props":29,"children":30},"p",{},[31],{"type":26,"value":32},"[Beijing, September 6, 2023] Baichuan AI held a conference in Beijing and officially released the Baichuan2 open-source foundation model, supported by the Ascend AI software and hardware platform. The Baichuan2-7B model is now freely available in the MindSpore community.",{"type":20,"tag":28,"props":34,"children":35},{},[36],{"type":26,"value":37},"At the launch event, the company announced the official open-source of fine-tuned models Baichuan2-7B, Baichuan2-13B, Baichuan2-13B-Chat, and a 4-bit quantitative version. These models, offering various services at no cost, cater to both academic and commercial markets and are ready for commercial deployment.",{"type":20,"tag":28,"props":39,"children":40},{},[41],{"type":20,"tag":42,"props":43,"children":45},"img",{"alt":7,"src":44},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2023/10/24/2aa6cd8667d347f2a8f89e0ba76392fa.png",[],{"type":20,"tag":28,"props":47,"children":48},{},[49],{"type":26,"value":50},"(Launch event of the Baichuan2 model)",{"type":20,"tag":28,"props":52,"children":53},{},[54],{"type":26,"value":55},"Link to the MindSpore open-source repository:",{"type":20,"tag":28,"props":57,"children":58},{},[59],{"type":20,"tag":60,"props":61,"children":65},"a",{"href":62,"rel":63},"https://gitee.com/mindspore/mindformers/blob/dev/research/baichuan2/baichuan2.md",[64],"nofollow",[66],{"type":26,"value":62},{"type":20,"tag":28,"props":68,"children":69},{},[70],{"type":26,"value":71},"Link to the foundation model platform in the MindSpore open-source community:",{"type":20,"tag":28,"props":73,"children":74},{},[75],{"type":20,"tag":60,"props":76,"children":79},{"href":77,"rel":78},"https://xihe.mindspore.cn/modelzoo/baichuan2_7b_chat",[64],[80],{"type":26,"value":77},{"type":20,"tag":28,"props":82,"children":83},{},[84],{"type":20,"tag":85,"props":86,"children":87},"strong",{},[88],{"type":26,"value":89},"Excellent Performance Surpassing LLaMA 2",{"type":20,"tag":28,"props":91,"children":92},{},[93],{"type":26,"value":94},"Baichuan2-7B-Base and Baichuan2-13B-Base, trained on 2.6 TB of high-quality multilingual data, have notably enhanced their proficiency in areas like mathematics, coding, security, logical reasoning, and semantic understanding. Simultaneously, they maintain the optimal features of prior open-source models, including content generation, smooth multi-turn dialogues, and easy deployment. When contrasted with the preceding 13B model, Baichuan2-13B-Base exhibits enhancements in various capabilities: a 49% improvement in mathematics, 46% in coding, 37% in security, 25% in logical reasoning, and 15% in semantic understanding.",{"type":20,"tag":28,"props":96,"children":97},{},[98],{"type":20,"tag":42,"props":99,"children":101},{"alt":7,"src":100},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2023/10/24/40959b4dee014c6cb2a21770c6b2322e.png",[],{"type":20,"tag":28,"props":103,"children":104},{},[105],{"type":26,"value":106},"The two open-source models demonstrate superior performance in key evaluation rankings. They outperform LLaMA 2 in authoritative benchmarks like MMLU, CMMLU, and GSM8K. Even when compared to other models with an equivalent number of parameters, they surpass the performance of other competing models.",{"type":20,"tag":28,"props":108,"children":109},{},[110],{"type":26,"value":111},"Notably, Baichuan2-7B, with seven billion parameters, performs on par with LLaMA 2, which has 13 billion parameters, in mainstream English tasks according to multiple authoritative benchmarks such as MMLU.",{"type":20,"tag":28,"props":113,"children":114},{},[115],{"type":20,"tag":42,"props":116,"children":118},{"alt":7,"src":117},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2023/10/24/982e5b3055d64918973f98b72a511cc9.png",[],{"type":20,"tag":28,"props":120,"children":121},{},[122],{"type":26,"value":123},"(Benchmark scores of models with 7 billion parameters)",{"type":20,"tag":28,"props":125,"children":126},{},[127],{"type":20,"tag":42,"props":128,"children":130},{"alt":7,"src":129},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2023/10/24/d1e2b0b98c1b4836a82e9e44dd9e01ae.png",[],{"type":20,"tag":28,"props":132,"children":133},{},[134],{"type":26,"value":135},"(Benchmark scores of models with 13 billion parameters)",{"type":20,"tag":28,"props":137,"children":138},{},[139],{"type":26,"value":140},"Baichuan2-7B and Baichuan2-13B not only are completely open to academic research, but also can be put into commercial use for free after developers apply for official commercial licenses by email.",{"type":20,"tag":28,"props":142,"children":143},{},[144],{"type":20,"tag":85,"props":145,"children":146},{},[147],{"type":26,"value":148},"Baichuan2 Foundation Model",{"type":20,"tag":28,"props":150,"children":151},{},[152],{"type":26,"value":153},"Baichuan2, both open-source and commercially available, is a series of pre-trained large language models developed by Baichuan AI. The series includes models with 7 billion, 13 billion, and 53 billion parameters. From its establishment, Baichuan AI has been committed to fostering the growth of China's foundational model industry via open sourcing. The two open-sourced Baichuan2 models have garnered favorable feedback from both upstream and downstream businesses. Numerous renowned companies, including Huawei, attended the launch event and entered into collaborations with Baichuan AI.",{"type":20,"tag":28,"props":155,"children":156},{},[157],{"type":20,"tag":85,"props":158,"children":159},{},[160],{"type":26,"value":161},"Ascend AI",{"type":20,"tag":28,"props":163,"children":164},{},[165],{"type":26,"value":166},"Ascend AI is an AI computing industry built based on the Ascend AI software and hardware platform. The platform includes Atlas series hardware and partner-branded hardware, heterogeneous computing architecture CANN, all-scenario AI framework MindSpore, Ascend application enablement MindX, one-stop development platform ModelArts, and unified toolchain MindStudio.",{"title":7,"searchDepth":168,"depth":168,"links":169},4,[],"markdown","content:news:en:2828.md","content","news/en/2828.md","news/en/2828","md",1776506046269]