diff --git a/docs/03-Deep Learning.ipynb b/docs/03-Deep Learning.ipynb
index e92914e..97aff1b 100644
--- a/docs/03-Deep Learning.ipynb	
+++ b/docs/03-Deep Learning.ipynb	
@@ -73,7 +73,7 @@
    "id": "0b9f2c9b-0ddc-4e07-bcd0-73b0ca876083",
    "metadata": {},
    "source": [
-    "As justified by our analysis of the reinforcement learnign for continuous control landscape (in the [paper](https://arxiv.org/abs/2306.03530)) in the beginning **RLtools** only supports fully connected neural networks. But we are planning on adding more architectures (especially recurrent neural networks) in the future.\n",
+    "As justified by our analysis of the reinforcement learning for continuous control landscape (in the [paper](https://arxiv.org/abs/2306.03530)) in the beginning **RLtools** only supports fully connected neural networks. But we are planning on adding more architectures (especially recurrent neural networks) in the future.\n",
     "\n",
     "We can instantiate a simple layer by first defining its hyperparameters (which are compile-time `constexpr` and types):"
    ]