kylechang523

After reading the code I found you change the max_pooling to max_pooling - avg_pooling, which get a difference. May I ask the reason doing this? I searched online but didn't...

Change dataset and caused NaN loss

7

Hi, while following the tutorial example in SimpleDnn, I changed the dataset which is after one-hot process, have 43 features, 11 labels. However I faced the problem shows 'Nan loss...

请问一下具体的训练细节

4

您好，您的模型效果非常好。 我最近也在开发crnn的相关模型，大致流程是先用自己生成的数据pretrain，再用真实数据finetuning，但是和您的结果比总是差了一些，想问一下： 1. 您生成的数据和真实数据按4:1 的比例，是混合在一起训练，还是pretrain+finetune？ 2. 您的模型有没有在SynthText上面pretrain过？ 十分期待您的回复，谢谢！ @wenlihaoyu

不知道如何调用predict_label来对模型的acc进行检查啊，希望能够予以解答！万分感谢！

May I ask the training time?

1

Hi, may I ask how many images do you use and how long will it cost for finish the training?

How to solve the unregistry bug during training process?

1

Hi, I want to re-implement the training pipeline and follow the README, but while loading the model I found this bug: KeyError: "No object named 'SOTR' found in 'META_ARCH' registry!"...

Training code

3

Hi, I'm super interested in this repo, is it possible to release the training code for learning purposes? Best, Kaiyu

Cannot install tensorRT==9.3.0

2

When I try to install tensorRT==9.3.0 for SD accelerate, I met this problem: ##### pip install tensorrt==9.3.0.post12.dev1 Defaulting to user installation because normal site-packages is not writeable Looking in indexes:...