# Periodic spikes in vf\_loss of PPO training

**URL:** <https://discuss.ray.io/t/periodic-spikes-in-vf-loss-of-ppo-training/4705>\
**Category:** RLlib\
**Created:** [January 14, 2022, 8:10pm UTC](https://discuss.ray.io/t/periodic-spikes-in-vf-loss-of-ppo-training/4705 "2022-01-14T20:10:10Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![vishalrangras](https://sea2.discourse-cdn.com/flex020/user_avatar/discuss.ray.io/vishalrangras/32/976_2.png) [@vishalrangras](https://discuss.ray.io/u/vishalrangras)\
**Post date:** [January 14, 2022, 8:10pm UTC](https://discuss.ray.io/t/periodic-spikes-in-vf-loss-of-ppo-training/4705/1 "2022-01-14T20:10:10Z")

</div>

@sven1977 What can I interpret from the periodic spikes in vf\_loss while training a PPO Policy? As you can see in the following screenshot, I keep getting periodic spikes even upto 100M timesteps. What does these spikes signifies? Is this the ideal representation of vf\_loss or the spikes should not be present?

 ![image](https://us1.discourse-cdn.com/flex020/uploads/ray/original/2X/e/e80f453374dec5a7ea84744f3c089122fd5c9cf4.png)