How should I save the loss value during training? model.fit has no return value

[Z-shuaishuai]

How should I save the loss value during training?
model.fit has no return value

[Darlanio]

I think you need to use Tensorflow.Keras.Engine.Sequential instead of model,
and then use predict, calculate loss from the result of predict,
calculate new gradients with Tensorflow.Gradients.GradientTape,
and finally use your keras.optamizers.[chosen optimizer].apply_gradients().

However, I am less than sure about this. I hope Haiping can answer this - possibly correcting my answer.

[Oceania2018]

model.fit returns History for next release v0.100.2.

[ammar-albush]

When will the new version(v0.100.2) be available on Nuget ? @Oceania2018

[Darlanio]

My guess, it will take at least a few days, probably a few weeks:
There are many small bugs to fix at present (927, 928... 973...)