search

chaneyddtt / Convolutional-Sequence-to-Sequence-Model-for-Human-Dynamics

Source code of paper "Convolutional Sequence to Sequence Model for Human Dynamics"
MIT License
86 stars 18 forks source link

Does trianing the model needs h5 file? #10

Open william-chen-zy opened 2 years ago

william-chen-zy commented 2 years ago

Hi, I am trying to train the model by doing python src/AC_main.py, however I run into the error OSError: Unable to create file (unable to open file: name = './samples/CNNAdTrain_GANWEIGHT0.010000_Sampling0.950000WindownLength20-2000.h5', errno = 2, error message = 'No such file or directory', flags = 15, o_flags = c2).

After reading rest of the issues I get confused about the .h5 file. Do I need to create it by running test first? or should this file be automatically generated when I run python src/AC_main.py?

Thanks

william-chen-zy commented 2 years ago

Add: in the Autoencoder_gan.py, why do we need to call inferenceSample in the Train method? Seems this is causing the problem during training.

image

chaneyddtt commented 2 years ago

Hi @william-chen-zy, we call the inferenceSample because we want to test our model every 2000 iterations. You might need to create a file named 'samples', where the saved model will be located in. Hope this helps.

william-chen-zy commented 2 years ago

Hi @chaneyddtt thank you for your prompt response. I commented out the self.InferenceSample line and was able to train the model successfully. My understanding is that in order to do Inference during the training (test model every 2000 iterations) I will need to have a trained model first. Am I right?

chaneyddtt commented 2 years ago

Yes, you can run the inference step after you have a trained model. Basically, the inference step is to compute error for each action and also save the predicted sequence for further visualization.