Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

请教:这个训练的loss为啥到1000步的时候突然收敛那么多? #9

Open
BrantonLiu opened this issue Nov 1, 2023 · 5 comments

Comments

@BrantonLiu
Copy link

training_loss
如题,在huggingface上看到train_loss这个图片,这个跑到1000步突然就降了很多是为啥呢?

@lemuria-wchen
Copy link
Contributor

猜测可能是 1000步骤左右刚好过了 1 个epoch

@lemuria-wchen
Copy link
Contributor

我们一共训练了2个epoch

@BrantonLiu
Copy link
Author

噢明白了 感谢~

@ccp123456789
Copy link

我们一共训练了2个epoch

这个loss趋势合理吗? 第一个epoch的时候阶梯式下降,是不是学习率什么的导致的?

@lemuria-wchen
Copy link
Contributor

我们一共训练了2个epoch

这个loss趋势合理吗? 第一个epoch的时候阶梯式下降,是不是学习率什么的导致的?

loss的阶梯式下降特别是在训练初期阶段应该是比较常见的现象。这代表模型正在从随机初始化LORA权重快速学习数据集的特征。对于大多数模型和优化算法(如SGD、Adam等)来说,第一个epoch结束时观察到显著的loss下降是合理的,因为模型参数开始适应训练数据。

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants