【量子位讯】近日，国内知名科技企业澜舟科技宣布，其自主研发的孟子3-13B大模型已正式开源，这一举措在科技界引起了广泛关注。孟子3-13B大模型以其高性价比和轻量化设计，成为学术研究领域的一大利器，并且开放了免费的商业使用权。
据澜舟科技官方介绍，这款大模型在训练过程中使用了万亿级别的token数据，确保了其强大的处理能力和广泛的适用性。在一系列权威基准测试如MMLU、GSM8K、HUMAN-EVAL中，孟子3-13B展现出不俗的性能，特别是在轻量化模型领域，当参数量控制在20B以内时，其在中英文语言处理能力上表现出色，同时在数学和编程能力上也位于行业前列。
这一开源决定不仅将促进学术界对大模型的深入研究，也有望推动人工智能技术在各行业的广泛应用。澜舟科技的这一创新举措，无疑为国内乃至全球的开发者和企业提供了宝贵的资源，将加速人工智能技术的普惠和创新步伐。
英语如下：
**News Title:** “LanZhou Tech Open Sources its Zhusuan 3-13B Large Language Model: Lightweight Efficiency Pioneers a New Era of AI for Free Commercial Use”
**Keywords:** LanZhou Tech, Zhusuan 3-13B, Large Model Open Source
**News Content:** _Quantum Bit News_ – Recently, renowned domestic tech firm LanZhou Tech announced that its self-developed Zhusuan 3-13B large language model has been officially open-sourced, drawing significant attention in the tech community. The Zhusuan 3-13B model, known for its high cost-performance ratio and lightweight design, has emerged as a powerful tool in academic research, and it offers free commercial usage rights.
According to LanZhou Tech’s official statement, the model was trained using trillions of tokens, guaranteeing its robust processing capabilities and wide applicability. It has demonstrated impressive performance in authoritative benchmarks like MMLU, GSM8K, and HUMAN-EVAL.特别是在轻量级模型领域，当参数量控制在20B以内时，Zhusuan 3-13B excels in both Chinese and English language processing tasks and ranks at the forefront in mathematical and programming abilities.
This open-source decision is poised to not only deepen academic research on large models but also facilitate the widespread adoption of AI technology across industries. LanZhou Tech’s innovative move provides invaluable resources to developers and businesses both domestically and globally, accelerating the democratization and innovation of AI technology.
【来源】https://mp.weixin.qq.com/s/L4wqVnbS8a9FT0Sd8JvDPw
Views: 1