Varun Gupta

Were you able to find a solution? Facing similar issues.

Found any solution?

Found it. Upon calling, `model.fit` you can store what the call return in a variable, and call `.history` on that object. Alternatively, you may dump it's contents in a pickle...

Did you find a solution? In my case, there is only `checkpoint` `_config.json` and checkpoint contains ``` model_checkpoint_path: ".00001" all_model_checkpoint_paths: ".00001" ```

Yes, I too then changed the code to save an '.h5' file, and now my model gets saved. (refered [this](https://stackoverflow.com/questions/70828564/tf-keras-callbacks-modelcheckpoint-callback-is-always-overwritten-despite-of-set#:~:text=from%20tensorflow.keras%20import%20callbacks%0A%0Acallbacks.ModelCheckpoint(%0A%20%20%20%20%20filepath%3D%27model.%7Bepoch%3A02d%7D%2D%7Bval_loss%3A.4f%7D.h5%27%2C%20%0A%20%20%20%20%20save_freq%3D%27epoch%27%2C%20verbose%3D1%2C%20monitor%3D%27val_loss%27%2C%20%0A%20%20%20%20%20save_weights_only%3DTrue%2C%20%0A) )) In your case, I atleast see index and data...

@sachinkmohan I tried doing this, but I'm still getting the error at `line 77, in get_vgg_encoder NameError: name 'keras' is not defined`. I also tried adding `from tensorflow import keras`....

@sachinkmohan Thank you for your inputs. I identified that the error though ending at this trace, was being caused by other issue, also I was running the command from subdir,...

@ssnirgudkar @SjoerdBraaksma I'm having some issues with the vgg model as well, may I know the tensorflow version you are using to train the model?

I do not have a CUDA 11 compatible GPU card. Thus, I have to use TF

Thanks for the reply @windj007. Changing `dataset.img_suffix=.jpg` at the given location did not resolve the issue. I still get the message: ``` - Make val dataloader default from /home2/varungupta/lama/ts_images/ 0it...