Module: tfa.optimizers.yogi

View source on GitHub

Yogi: Extension of yogi adaptive nonconvex optimizer in Keras.

Implementation of Additive Averaging. m_t+1 = beta1m_t + (1-beta1)g_t v_t+1 = v_t + sign(g_t-v_t)(g_t^2) Experiments show better performance across NLP and Vision tasks. Paper: https://papers.nips.cc/paper/8186-adaptive-methods-for-nonconvex-optimization.pdf

Classes

class Yogi: Optimizer that implements the Yogi algorithm in Keras.

Type Aliases

FloatTensorLike