You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
------------------ 原始邮件 ------------------
发件人: ***@***.***>;
发送时间: 2025年1月7日(星期二) 中午11:15
收件人: ***@***.***>;
抄送: ***@***.***>; ***@***.***>;
主题: Re: [QwenLM/Qwen2-VL] Qwen2-VL-72B-Instrcut (Issue #637)
一个是chat model,是base模型经过alignment用于chat,会有一定的偏好,base模型是预训练模型主要完成续写,在指令跟随会差些,可以根据自己的场景进行调整
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
请问Qwen2-VL-72B-Instrcut和Qwen2-VL-72B-Base的差异大吗,如何根据场景选择更适合自己的模型?
The text was updated successfully, but these errors were encountered: