分类准确率下降，性能提升一倍，咋解决呢？

sgl-project / sglang

SGLang is a fast serving framework for large language models and vision language models.

https://sgl-project.github.io/

Apache License 2.0

5.8k stars 464 forks source link

Closed xxm1668 closed 3 months ago

xxm1668 commented 8 months ago

qwen自带推理：第一身份准确率：0.968421052631579 第二身份准确率：0.9368421052631579 一条数据平均耗时：0.9585727064233077

slang推理：第一身份准确率：0.6631578947368421 第二身份准确率：0.7894736842105263 一条数据平均耗时：0.5773725032806396

Ying1123 commented 8 months ago

@xxm1668 你好，可以给一些例子，使用方法信息，让我们复现吗？可能是prompt没有对齐之类的原因。

xxm1668 commented 8 months ago

微调的qwen1.5-chat-14b，对电话录音文本说话人进行分类的。输出的结果如：{"对话人1身份":"","对话人2身份":""}

github-actions[bot] commented 3 months ago

This issue has been automatically closed due to inactivity. Please feel free to reopen it if needed.