Update README.md
Browse files
README.md
CHANGED
@@ -32,4 +32,33 @@ tags:
|
|
32 |
2. Step 1: Find your model_id: chirbard/ppo-Worm
|
33 |
3. Step 2: Select your *.nn /*.onnx file
|
34 |
4. Click on Watch the agent play 👀
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
35 |
|
|
|
32 |
2. Step 1: Find your model_id: chirbard/ppo-Worm
|
33 |
3. Step 2: Select your *.nn /*.onnx file
|
34 |
4. Click on Watch the agent play 👀
|
35 |
+
|
36 |
+
## Hyperparameters
|
37 |
+
```
|
38 |
+
behaviors:
|
39 |
+
Worm:
|
40 |
+
trainer_type: ppo
|
41 |
+
hyperparameters:
|
42 |
+
batch_size: 2024
|
43 |
+
buffer_size: 20240
|
44 |
+
learning_rate: 0.0003
|
45 |
+
beta: 0.005
|
46 |
+
epsilon: 0.2
|
47 |
+
lambd: 0.95
|
48 |
+
num_epoch: 3
|
49 |
+
learning_rate_schedule: linear
|
50 |
+
network_settings:
|
51 |
+
normalize: true
|
52 |
+
hidden_units: 512
|
53 |
+
num_layers: 3
|
54 |
+
vis_encode_type: simple
|
55 |
+
reward_signals:
|
56 |
+
extrinsic:
|
57 |
+
gamma: 0.9995
|
58 |
+
strength: 1.0
|
59 |
+
keep_checkpoints: 5
|
60 |
+
max_steps: 5000000
|
61 |
+
time_horizon: 1000
|
62 |
+
summary_freq: 30000
|
63 |
+
```
|
64 |
|