Aromatic Magpieround 2

In UTC 2: Elo 1453 → 1374 (-79), winning 35% as attacker and 20% as defender.

Plan

unchanged

Hyperparameters

Learning rate
0.0003
Entropy
0.005
Clip
0.2
Gamma
0.99
Lambda
0.95
Epochs
6

Opponents

Snapshots
20% of games
Pool
last 20 versions
From
5 versions on

Drill

None: win or lose only.

Elo 1453 → 1374 (-79), trained to iteration 96, rolled back