Apr 16, 04:07 PM

CodeQwen1.5-7B Models Outperform Rivals, Offer Open Access

The new CodeQwen1.5-7B series, including CodeQwen1.5-7B and CodeQwen1.5-7B-Chat, has been introduced, featuring advancements in language model technology for coding tasks. These models are built on the Qwen1.5 language model, pre-trained with 3 trillion tokens of code-related data. The CodeQwen1.5-7B model supports various tasks such as code generation, code editing, SQL, and chat functionalities. It is noted for its performance, outperforming competitors like DeepSeek Coder and ChatGPT 3.5 on the SWE bench. Additionally, the models are open access with weights available on the Hub and require about 5GB of RAM.

#Code #Chat #SQL #DeepSeek Coder #ChatGPT #SWE #Hub

Written with ChatGPT (GPT-4).