Weihan Li

如题 现在有多张图片,需要合并为一个tensor 是应该转换为一个三维Mat(包含多张图片),再使用相关接口(如果有的话)转换为对应维度的tensor 还是应该逐个图片(二维Mat)转换为对应Tensor,再使用相关接口合并低维度tensor为高维度tensor

我已经修改了SamRunner里面的在.cuda时候的gpu_id,以及get_model时候的gpu_id,仍然提示我有一部分参数or数据在cuda:0上 我应当如何解决

About training environment！！！

4

Hello there, I've been suffered from the training environment about 3days just because of using a shared server in my lab. Here's sth you need to notice while you're config...

About cogvlm

1

Hello, I'm using your VIE score to assess my own diffusion model, it helps me a lot. But when I turn to use cogvlm to test my model, I found...