diff --git a/docs/10.md b/docs/10.md index c38bee40ba99d87ab91d833bf34c9ececb6b2875..91fe8eea467bbf6cdee05c99fbcef382a3aff65a 100644 --- a/docs/10.md +++ b/docs/10.md @@ -554,6 +554,8 @@ $$ **练习 6.7** 这里是对离散动作空间使用自动微分来执行最大似然估计的伪代码。 -${\sf logits = policy.predictions(states)}$ +$\text{logits = policy.predictions(states)}$ -${\sf negative_likelihoods = tf.nn.softmax_cross_entropy_with_logits(labels=actions, logits=logits)}$ \ No newline at end of file +$\text{negative_likelihoods = tf.nn.softmax_cross_entropy_with_logits(}$ + +$\text{labels=actions, logits=logits)}$ \ No newline at end of file