如何用多模态模型做图像分类任务? #175
wesleyliwei
started this conversation in
General
Replies: 2 comments
-
做了一些尝试,比如只finetune vit后面的MLP层,整体分类任务的train和val loss都在下降,但是最后预测test数据集的yes和no的效果其实没有那么好。 PS:训练数据的label只是yes和no。 test分类label也是yes和no |
Beta Was this translation helpful? Give feedback.
0 replies
-
可以直接用语言模型微调。比如做图片做狗和猫的分类。准备sft数据,语言模型的reponse就两种,一个"狗",一个"猫"就行。然后用这个数据和微调代码微调就行。 |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
大家好~ 不知道大家有没有用cogvlm2做图像分类训练的,如果是一个类似图像分类任务的话,finetune大家建议哪些params设置为可惜训练的呢?
Beta Was this translation helpful? Give feedback.
All reactions