diff --git a/README.md b/README.md
index 2ddd00c0c..9a2f65223 100644
--- a/README.md
+++ b/README.md
@@ -15,7 +15,7 @@ full and half WPA handshakes.
 
 ![ui](https://i.imgur.com/X68GXrn.png)
 
-Instead of merely playing [Super Mario or Atari games](https://becominghuman.ai/getting-mario-back-into-the-gym-setting-up-super-mario-bros-in-openais-gym-8e39a96c1e41?gi=c4b66c3d5ced) like most reinforcement learning-based "AI" *(yawn)*, Pwnagotchi tunes [its parameters](https://github.com/evilsocket/pwnagotchi/blob/master/pwnagotchi/defaults.yml#L73) over time to **get better at pwning WiFi things to** in the environments you expose it to. 
+Instead of merely playing [Super Mario or Atari games](https://becominghuman.ai/getting-mario-back-into-the-gym-setting-up-super-mario-bros-in-openais-gym-8e39a96c1e41?gi=c4b66c3d5ced) like most reinforcement learning-based "AI" *(yawn)*, Pwnagotchi tunes [its parameters](https://github.com/evilsocket/pwnagotchi/blob/master/pwnagotchi/defaults.toml) over time to **get better at pwning WiFi things to** in the environments you expose it to. 
 
 More specifically, Pwnagotchi is using an [LSTM with MLP feature extractor](https://stable-baselines.readthedocs.io/en/master/modules/policies.html#stable_baselines.common.policies.MlpLstmPolicy) as its policy network for the [A2C agent](https://stable-baselines.readthedocs.io/en/master/modules/a2c.html). If you're unfamiliar with A2C, here is [a very good introductory explanation](https://hackernoon.com/intuitive-rl-intro-to-advantage-actor-critic-a2c-4ff545978752) (in comic form!) of the basic principles behind how Pwnagotchi learns. (You can read more about how Pwnagotchi learns in the [Usage](https://www.pwnagotchi.ai/usage/#training-the-ai) doc.)