New content on http://keras.io : Keras debugging tips.https://keras.io/examples/keras_recipes/debugging_tips/ …
Show this thread
-
1 - Test the parts before you test the whole 2 - Use `model.summary()` and `plot_model()` to check output shapes and connectivity graph 3 - Use `run_eagerly=True` in `compile()` to debug training step-by-step 4 - Use the TF profiler (simple callback) to fix performance issues
Show this thread
Also, yes -- if you ever find yourself needing second-order gradients when writing a Keras training step, this is how to get them:pic.twitter.com/FY7fm8eE2k
-
-
I always wonder when/where to use the
@tf.function decorator. In your example provided here, would it be useful to use it to decorate the train_step function ? -
If you're using fit/predict, you never need to use tf.function (just pass run_eagerly=True/False in compile). If you're writing low-level training loops from scratch, add a tf.function decorator to your training step (just one in total).
End of conversation
New conversation -
-
-
@threader_app please compile it. -
Hey, thanks for sharing this thread from
@fchollet. Read the whole version here:https://threader.app/thread/1261768958153076736 …
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.