背景
续GPT爆火后,大模型相关论文研究开始突飞猛进增长
![image-20240205150022524](https://i0.wp.com/imgs.run/i/2024/02/10/65c719f8bab16.png?w=640&ssl=1)
研究方向也逐步由专用模型(针对特定任务,一个模型解决一个问题)转变为使用通用大模型(一个模型应对多个任务及模态)
![image-20240205150526522](https://i0.wp.com/imgs.run/i/2024/02/10/65c71a08388ef.png?w=640&ssl=1)
书生·浦语大模型
书生·浦语大模型提供了不同量级的大模型版本:
轻量级:InternLM-7B
中量级:InternLM-20B
重量级:InternLM-123B
![image-20240205150513147](https://i0.wp.com/imgs.run/i/2024/02/10/65c71a12640ef.png?w=640&ssl=1)
书生·浦语模型性能如下:
![image-20240205150604448](https://i0.wp.com/imgs.run/i/2024/02/10/65c71a1ca4816.png?w=640&ssl=1)
从模型到应用
![image-20240205161124966](https://i0.wp.com/imgs.run/i/2024/02/10/65c71a2567648.png?w=640&ssl=1)
相关开放体系
![image-20240205161310903](https://i0.wp.com/imgs.run/i/2024/02/10/65c71a2eee59d.png?w=640&ssl=1)
![image-20240205161402915](https://i0.wp.com/imgs.run/i/2024/02/10/65c71a393e2cb.png?w=640&ssl=1)
![image-20240205161415481](https://i0.wp.com/imgs.run/i/2024/02/10/65c71a4326abc.png?w=640&ssl=1)
![image-20240205161422793](https://i0.wp.com/imgs.run/i/2024/02/10/65c71a4d23b65.png?w=640&ssl=1)
![image-20240205161432269](https://i0.wp.com/imgs.run/i/2024/02/10/65c71a562ffa8.png?w=640&ssl=1)
![image-20240205161439752](https://i0.wp.com/imgs.run/i/2024/02/10/65c71a608c1a7.png?w=640&ssl=1)
![image-20240205161449144](https://i0.wp.com/imgs.run/i/2024/02/10/65c71a6b18cf7.png?w=640&ssl=1)
![image-20240205161505281](https://i0.wp.com/imgs.run/i/2024/02/10/65c71a7557104.png?w=640&ssl=1)
![image-20240205161512483](https://i0.wp.com/imgs.run/i/2024/02/10/65c71a7e0ed95.png?w=640&ssl=1)
![image-20240205161524527](https://i0.wp.com/imgs.run/i/2024/02/10/65c71a8859fcc.png?w=640&ssl=1)
![image-20240205161603019](https://i0.wp.com/imgs.run/i/2024/02/10/65c71a9237a7d.png?w=640&ssl=1)
![image-20240205162925294](https://i0.wp.com/imgs.run/i/2024/02/10/65c71a9c3cde9.png?w=640&ssl=1)
![image-20240205163018266](https://i0.wp.com/imgs.run/i/2024/02/10/65c71aa5f1b93.png?w=640&ssl=1)
![image-20240205163046390](https://i0.wp.com/imgs.run/i/2024/02/10/65c71aaf344c3.png?w=640&ssl=1)
![image-20240205163144277](https://i0.wp.com/imgs.run/i/2024/02/10/65c71ab93b970.png?w=640&ssl=1)
![image-20240205163249597](https://i0.wp.com/imgs.run/i/2024/02/10/65c71ac38eac5.png?w=640&ssl=1)
4dBmk·2024-02-09·6 次阅读
续GPT爆火后,大模型相关论文研究开始突飞猛进增长
研究方向也逐步由专用模型(针对特定任务,一个模型解决一个问题)转变为使用通用大模型(一个模型应对多个任务及模态)
书生·浦语大模型提供了不同量级的大模型版本:
轻量级:InternLM-7B
中量级:InternLM-20B
重量级:InternLM-123B
书生·浦语模型性能如下:
Comments NOTHING