Sheep training flock _ improver

2026-04-24 23:27:05 +01:00
parent b3251fcca3
commit 0716c6c3c8
2 changed files with 2 additions and 2 deletions
@@ -110,7 +110,7 @@ def train_stage(n_sheep, steps, n_envs=4, prev_model=None, prev_vecnorm=None):
        model = PPO(
            "MlpPolicy", vn,
            learning_rate=3e-4, n_steps=2048, batch_size=256, n_epochs=10,
-            gamma=0.995, gae_lambda=0.95, clip_range=0.2, ent_coef=0.02,
+            gamma=0.995, gae_lambda=0.95, clip_range=0.2, ent_coef=0.005,
            vf_coef=0.5, max_grad_norm=0.5,
            policy_kwargs=dict(net_arch=[256, 256]),
            verbose=1,
@@ -332,7 +332,7 @@ def main():
        gamma           = 0.995,
        gae_lambda      = 0.95,
        clip_range      = 0.2,
-        ent_coef        = 0.02,
+        ent_coef        = 0.005,
        vf_coef         = 0.5,
        max_grad_norm   = 0.5,
        policy_kwargs   = dict(net_arch=[256, 256]),