Spaces:
Running
Running
Added screenshots and text
Browse files- .gitignore +2 -1
- assets/eval_screenshot_train_20k.png +0 -0
- assets/eval_screenshot_train_2k.png +0 -0
- index.html +49 -0
.gitignore
CHANGED
|
@@ -1,3 +1,4 @@
|
|
| 1 |
.env
|
| 2 |
frozen_lake_env
|
| 3 |
-
__pycache__
|
|
|
|
|
|
| 1 |
.env
|
| 2 |
frozen_lake_env
|
| 3 |
+
__pycache__
|
| 4 |
+
node_modules
|
assets/eval_screenshot_train_20k.png
ADDED
|
assets/eval_screenshot_train_2k.png
ADDED
|
index.html
ADDED
|
@@ -0,0 +1,49 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
<!DOCTYPE html>
|
| 2 |
+
<html>
|
| 3 |
+
|
| 4 |
+
<head>
|
| 5 |
+
<meta charset="utf-8">
|
| 6 |
+
<title>Frozen Lake</title>
|
| 7 |
+
<meta name="viewport" content="width=device-width, initial-scale=1">
|
| 8 |
+
<style>
|
| 9 |
+
body {
|
| 10 |
+
background-color: #000;
|
| 11 |
+
}
|
| 12 |
+
|
| 13 |
+
#container {
|
| 14 |
+
margin: auto;
|
| 15 |
+
max-width: 800px;
|
| 16 |
+
text-align: center;
|
| 17 |
+
}
|
| 18 |
+
|
| 19 |
+
#container>img {
|
| 20 |
+
width: 100%
|
| 21 |
+
}
|
| 22 |
+
|
| 23 |
+
#container>a,
|
| 24 |
+
#container>h2,
|
| 25 |
+
#container>p {
|
| 26 |
+
color: #fff;
|
| 27 |
+
}
|
| 28 |
+
|
| 29 |
+
#container>a {
|
| 30 |
+
margin-top: 16px;
|
| 31 |
+
}
|
| 32 |
+
</style>
|
| 33 |
+
</head>
|
| 34 |
+
|
| 35 |
+
<body>
|
| 36 |
+
<div id="container">
|
| 37 |
+
<h2>RL - Slippery Frozen Lake Q-Learning</h2>
|
| 38 |
+
<p>I trained a Q-Learning model on the OpenAI Gym Slippery Frozen Lake environment for 20,000 iterations, and
|
| 39 |
+
evaluated for 1,000 iterations.
|
| 40 |
+
The trained model had a success rate of about 73%. Action for the 3 column on the second row is especially
|
| 41 |
+
interesting,
|
| 42 |
+
it correctly learns that the best policy is to try move toward one of the holes because there is a 1/3
|
| 43 |
+
chance of slipping and slips are orthogonal to the desired direction.</p>
|
| 44 |
+
<img src="assets/eval_screenshot_train_20k.png" alt="Q-Learning Agent">
|
| 45 |
+
<a href="https://www.youtube.com/watch?v=b1oh3TK6Jhg">Training and Evaluation Video (youtube)</a>
|
| 46 |
+
</div>
|
| 47 |
+
</body>
|
| 48 |
+
|
| 49 |
+
</html>
|