何森 comments

Results 10 comments of


                                            何森

nan loss

my batchsize is 4, I use single GPU to train the model. Even in CelebA dataset, I use your original code to create the lmdb data and then train the...

nan loss

I increased the --batch from 4 to 8, decreased --lr from 0.002 to 0.0002 and set --g_reg_every to be larger than the total training iterations. The loss still become NaN...

训练检测点

Please check the in the [code](https://github.com/SenHe/Flow-Style-VTON/blob/cbc0bdf84d9b6877d916f625052cb108ebd881b5/train/train_PBAFN_stage1_fs.py#L184)

训练检测点

Please make sure you have read and understood all the code and then ask this question.

Python built-in function for extracting the cloth's edge image

please refer to [this issue](https://github.com/SenHe/Flow-Style-VTON/issues/6).

Hello, I have a problem that the calculated output size is too small (51210）when I train the model, is there anyone who have successfully run the pretrained train_PBAFN_stage1_fs.sh model could give me some suggestions?

please check your input size, the PyTorch version you are using, etc.

Hello, I have a problem that the calculated output size is too small (51210）when I train the model, is there anyone who have successfully run the pretrained train_PBAFN_stage1_fs.sh model could give me some suggestions?

please refer to a [previous issue](https://github.com/SenHe/Flow-Style-VTON/issues/12).

flow visualization codes

Please check [here](https://github.com/SenHe/Flow-Style-VTON/blob/1e5df26abab4ef1fce9ec75cc27b250c011fe639/test/test.py#L111). You can amend to visualize the sampling grid or the offset.

get bad results

Please refer to previous issues [#11](https://github.com/SenHe/Flow-Style-VTON/issues/11), [#3](https://github.com/SenHe/Flow-Style-VTON/issues/3)

Evaluation on official trained model

> Hello, thanks a lot for opening your source code. Now I am reimplementing your work. But the metrics from the trained model are not the same as the papers....