search

herobd / handwriting_line_generation

Code for BMVC2020 paper "Text and Style Conditioned GAN for Generation of Offline Handwriting Lines"
Other
66 stars 29 forks source link

’self._ valid_ Epoch() 'Wrong problem #36

Open Sonel-YXL opened 1 year ago

Sonel-YXL commented 1 year ago

This is a great job When I first started encoder training. In line 279 of "basetrainer. py", there is a problem with "val result=self. _validd_Epoch". This function cannot be found. My current solution is to annotate valid modules. Is there a better way to solve this problem? Thank you.

herobd commented 1 year ago

It looks like a typo was introduced to your copy of the code, it is val_result = self._valid_epoch() in the repo

Sonel-YXL commented 1 year ago

Sorry, I didn't describe this problem very clearly. The following is the original error message.

Train iteration: 900, loss: 3.044715, recogLoss: 3.044715, CER: 0.990476, WER: 1.000000, sec_per_iter: 0.320079, avg_loss: 3.146585, avg_recogLoss: 3.146585, avg_CER: 0.997909, avg_WER: 1.000000,
validate Train iteration: 1000, loss: 2.888014, recogLoss: 2.888014, CER: 0.979681, WER: 1.000000, sec_per_iter: 0.319059, avg_loss: 3.046097, avg_recogLoss: 3.046097, avg_CER: 0.990086, avg_WER: 0.999848,
WARNING: upsampling image to fit size Traceback (most recent call last): File "/home/sonel/code/dormitory_code_20221104/GAN/handwriting_line_generation-master/train.py", line 132, in main(config, args.resume) File "/home/sonel/code/dormitory_code_20221104/GAN/handwriting_line_generation-master/train.py", line 78, in main trainer.train() File "/home/sonel/code/dormitory_code_20221104/GAN/handwriting_line_generation-master/base/base_trainer.py", line 279, in train val_result = self._valid_epoch() File "/home/sonel/code/dormitory_code_20221104/GAN/handwriting_line_generation-master/trainer/hw_with_style_trainer.py", line 464, in _valid_epoch pred, recon, losses = self.run(instance) AttributeError: 'HWWithStyleTrainer' object has no attribute 'run'

herobd commented 1 year ago

What is your training config?

Sonel-YXL commented 1 year ago

My training configuration is '-c configs/cf_IAM_hwr_cnnOnly_batchnorm_aug.json' In the json file, I only changed the file_path. I initially queried whether there is val_result = self._valid_epoch() function in the class, but I didn't seem to find it.So I'm considering whether it is caused by this factor. Because this is the verification part, I deleted all the verification parts without affecting the final training. There must be a better solution here. Thank you for your detailed answer,have a good day！

herobd commented 1 year ago

Looks like a bug from a refactor. Line 464 should be pred, recon, losses = self.run_hwr(instance) I've now fixed this in the repo. Change your code and try it.

Sonel-YXL commented 1 year ago

Thank you for your advice, I will have a try