Tensorflow Federated: given a model train it with 2 different datsets and calcaulate euclidian distance between these models

Question

I have a next word prediction model based on federated learning with tensorflow model. My server need to calculate the distance between the model weigtht I receive in each round. Do you have any idea how to do it?

user11530462 · Accepted Answer · 2022-12-01 17:23:00Z

0

Get the model weights using model.get_weights() for each model,

Then make them as a tensor. For example

model1_weights= model1.get_weights()
model2_weights= model2.get_weights()

weights= tf.constant([[model1_weights],[model2_weights]])

Then you can find the euclidean distance using

tf.math.reduce_euclidean_norm(x)

For more details you can refer to this gist. Thank You.

answered Dec 1, 2022 at 17:23

user11530462

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Tensorflow Federated: given a model train it with 2 different datsets and calcaulate euclidian distance between these models

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related