CaiLiYuan

Results 8 comments of CaiLiYuan

Error during resuming training when i tried resuming from last saved checkpoints.

same problem！Has anyone solved it?

Resume training fails with 'NoneType' object is not subscriptable

same problem! The speed of training is very slow and It takes a lot of time to retrain。

RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu!

@zengjie617789 good job!

Multiple GPUs Problem

#296

为什么最新的拉下来训练的很慢？ 比半个月拉下来额慢了好几本 配置一样

对比V7,V5,V3的代码，V6的训练时间特别慢

mmdet.FocalLoss training error

when I use mmdet.DistributionFocalLoss or mmdet.GaussianFocalLoss ,bug is ``` # mmdet.DistributionFocalLoss or mmdet.GaussianFocalLoss bug 2/05 23:06:20 - mmengine - INFO - Result has been saved to /home/xux/CaiLiYuan/Project/mmyolo/work_dirs/yolov7_x_Dfl/modules_statistic_results.json Traceback (most recent...

infer.py精度与eval.py精度差距过大

可能eval开启了--reproduce_640_eval，而infer没开启，这个参数会执行包含了letterbox的操作，你可以测试下宽高比为1的图片测试下eval和infer的区别

How to use dpm for guided diffision model(ddpm)

I directly replace the DDPM sampling of the Palette model with dpmsolver sampling. But the results obtained are very poor. Does this require retraining? I use the same code with...