a3fe8bdf1c484e390dfe3947cc395372c0187589,tensorforce/core/distributions/beta.py,Beta,tf_parametrize,#Beta#Any#,50

Before Change

        beta = self.beta.apply(x=x)
        beta = tf.clip_by_value(t=beta, clip_value_min=log_eps, clip_value_max=-log_eps)
        beta = tf.log(x=(tf.exp(x=beta) + 1.0)) + 1.0

        shape = (-1,) + self.action_spec["shape"]
        alpha = tf.reshape(tensor=alpha, shape=shape)

After Change

        beta = self.beta.apply(x=x)
        // epsilon < 1.0, hence negative
        beta = tf.clip_by_value(t=beta, clip_value_min=log_epsilon, clip_value_max=-log_epsilon)
        beta = tf.math.softplus(features=beta) + one
        beta = tf.reshape(tensor=beta, shape=shape)

        // Alpha + Beta

Instances