Haitao Lin

I think DDPM is a variation preserving process and DSM is the exploring one. When your method is variation preserving one, the `DIFFUSION` may not be a suitable description. The...

Hi, Minkai. I also agree with you that the time embedding is not necessary to used for calculating the noise. Thanks for your detailed explanation. I am also thinking the...

At first, I guess that the second global encoder makes the code invalid since the rescaling problem with the cutoff hyperparameter. And I use a single encoder version where the...

实验部分有单位介绍，为了将单位的小数位数统一，因此表格中的单位并非实际单位。

一旦使用滑动，就会出现标签泄漏。即需要预测的输出会在下一次作为输入，被模型所见。 因此这里直接采用按24小时分割，避免标签泄漏

你好，我们平台还在集成搭建中。 我后续的一篇与lirong wu合作的文章为promptDDG，比这篇的相关精确度更高，已经开源。如果需要使用，我们也优先上线那个。您可以直接去相关codebase下载。 谢谢。

The newly trained ppflow as `pretrained.pt` from [Google Drive](https://drive.google.com/drive/u/0/folders/1ce5DVmZz0c-p3PKrGDQoU_C9MD3cWLNq) with the data provided from [PepGlad](https://github.com/THUNLP-MT/PepGLAD/tree/main) is uploaded. The performance is different from the reported ones since the training data is...

你好 暂且还没有开放相关功能。From Lin Haitao’s iphone.在 2024年7月25日，09:15，Dadashuai1042 ***@***.***> 写道： 非常感谢您的开源，很优秀的工作，但是在运行代码时，我有一个疑问，程序的输入输出是什么，说明文档的运行步骤中只有train和evaluation，能够根据自己输入的抗原生成抗体吗 —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you are subscribed to this thread.Message...