jalf/TIR_PROJ
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Sheep training flock of 10 fix?

Browse Source
This commit is contained in:
Johnny Fernandes
2026-04-24 15:14:45 +01:00
parent bdbe8ba1de
commit 44b2788e78
1 changed files with 4 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files
training/smoke_test.py
+4 -6
View File
@@ -98,11 +98,8 @@ def train_stage(n_sheep, steps, n_envs=4, prev_model=None, prev_vecnorm=None):
vn = VecNormalize(train_env, norm_obs=True, norm_reward=True, clip_obs=10.0) vn = VecNormalize(train_env, norm_obs=True, norm_reward=True, clip_obs=10.0)
if prev_model is not None: if prev_model is not None:
model = PPO.load(prev_model, env=vn, model = prev_model
learning_rate=3e-4, n_steps=2048, batch_size=256, model.set_env(vn)
n_epochs=10, gamma=0.995, gae_lambda=0.95,
clip_range=0.2, ent_coef=0.005, vf_coef=0.5,
max_grad_norm=0.5)
else: else:
model = PPO( model = PPO(
"MlpPolicy", vn, "MlpPolicy", vn,
@@ -113,7 +110,8 @@ def train_stage(n_sheep, steps, n_envs=4, prev_model=None, prev_vecnorm=None):
verbose=1, verbose=1,
) )
model.learn(total_timesteps=steps, reset_num_timesteps=(prev_model is None)) model.learn(total_timesteps=steps, reset_num_timesteps=(prev_model is None),
tb_log_name="ppo_smoke")
return model, vn return model, vn