Ootzk commited on
Commit
ba151de
·
verified ·
1 Parent(s): c0d15c1

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +43 -0
README.md ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: mit
3
+ tags:
4
+ - reinforcement-learning
5
+ - onnx
6
+ - pikachu-volleyball
7
+ library_name: stable-baselines3
8
+ ---
9
+
10
+ # alphachu-v1
11
+
12
+ First deployed model for [world-tournament](https://github.com/alphachu-volleyball/world-tournament).
13
+
14
+ ## Source
15
+
16
+ - **Experiment**: [016_continue_015_extend](https://github.com/orgs/alphachu-volleyball/projects/1?pane=issue&itemId=172792980), checkpoint_000015 (step ~111M)
17
+ - **W&B run**: https://wandb.ai/ootzk/alphachu-volleyball/runs/tv00fzku
18
+ - **Training**: curriculum crossplay, pre-duckll:5 unlock
19
+
20
+ ## Model config
21
+
22
+ | Field | Value |
23
+ |-------|-------|
24
+ | Side | player_1 |
25
+ | Action simplified | true |
26
+ | Obs simplified | false |
27
+ | Obs normalized | true |
28
+ | Size | 29.7 KB (ONNX) |
29
+
30
+ ## Win rates at checkpoint_000015 (Iter 15, 100 games each)
31
+
32
+ | Opponent | Win | Loss | Win rate |
33
+ |----------|-----|------|----------|
34
+ | random | 100 | 0 | 100% |
35
+ | stone | 100 | 0 | 100% |
36
+ | builtin | 100 | 0 | 100% |
37
+ | duckll:0 | 100 | 0 | 100% |
38
+ | duckll:1 | 100 | 0 | 100% |
39
+ | duckll:2 | 94 | 6 | 94% |
40
+ | duckll:3 | 90 | 10 | 90% |
41
+ | duckll:7 | 85 | 15 | 85% |
42
+
43
+ Average pool win rate: 96.1%