In your opinion, what's a feature that Keras is missing, or otherwise something that Keras does poorly at this time? Be extremely specific (precise workflow, API signature). Thanks!
Show this thread
-
Being able to use different learning rates for different layers w/o gradient tape. (I like PyTorch’s implementation of this feature)
You should be able to do this straightforwardly in a custom train_step, just use one optimizer per layer, iterate over the layers and do e.g. `opt[i]apply_gradients(zip(grads[slice_i], layers[i].trainable_weights))`
