Latest Search
Quote
| Back Zoom + Zoom - | ||||
|
Alibaba Cloud Launches Open-Source Qwen 2.5 with 100+ Models
Recommend 25 Positive 31 Negative 8 |
||||
|
||||
|
Zhou Jingren, Chief Technology Officer of Alibaba Cloud, unveiled Qwen 2.5, a new-generation open-source model of Tongyi Qianwen, at the 2024 Apsara Conference. The full series includes large language models, multimodal models, mathematical models, and code models in various sizes. Each size has a base version, an instruction-following version, and a quantized version, totaling over 100 models. Among them, the flagship model Qwen 2.5-72B outperforms Llama 405B. Compared to Qwen2, the entire Qwen2.5 model series are pre-trained on 18T tokens data, with an overall performance improvement of more than 18%, paired with more knowledge, stronger programming and mathematical abilities. The Qwen2.5-72B model scored 86.8 on the MMLU-rudex benchmark (examining general knowledge), 88.2 on the MBPP benchmark (examining code skills) and MATH (mathematical skills) benchmarks. Qwen 2.5 can generate up to 8,000 characters of content and supports more than 29 languages, including Chinese, English, French, Spanish, Russian, Japanese, Vietnamese and Arabic. AAStocks Financial News |
||||
Copied to Clipboard
 |
||||
