From 7c082d53e51943675b3821e16eb992470797bced Mon Sep 17 00:00:00 2001 From: Katherine Crowson Date: Sat, 13 Oct 2018 16:00:33 -0700 Subject: [PATCH] Fix Padam unbiased learning rate --- keras_contrib/optimizers/padam.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/keras_contrib/optimizers/padam.py b/keras_contrib/optimizers/padam.py index 9786b7194..77615129b 100644 --- a/keras_contrib/optimizers/padam.py +++ b/keras_contrib/optimizers/padam.py @@ -54,7 +54,7 @@ def get_updates(self, loss, params): K.dtype(self.decay)))) t = K.cast(self.iterations, K.floatx()) + 1 - lr_t = lr * (K.sqrt(1. - K.pow(self.beta_2, t)) / + lr_t = lr * (K.pow(1. - K.pow(self.beta_2, t), self.partial) / (1. - K.pow(self.beta_1, t))) ms = [K.zeros(K.int_shape(p), dtype=K.dtype(p)) for p in params]