jcreek/CosmicClash
1
0
Fork 0
mirror of https://github.com/jcreek/CosmicClash.git synced 2026-09-13 15:52:04 +00:00
Code Issues Packages Projects Releases Wiki Activity

feat(*): Add bot-vs-bot Spectate mode with main-menu entry, entropy-control flags (--ent-coef, --reset-std) for resumed training runs, and a Linux/3090 remote-training guide (TRAINING_LINUX.md)

Browse Source
This commit is contained in:
Josh Creek
2026-07-19 10:10:32 +01:00
parent 379ef9910e
commit 07217c3517
8 changed files with 230 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
training/train.py
+18 -3
View File
@@ -39,6 +39,13 @@ def parse_args():
parser.add_argument("--port", type=int, default=11008, help="Base TCP port (one per instance)")
parser.add_argument("--seed", type=int, default=0)
parser.add_argument("--resume", default=None, help="Checkpoint .zip to resume from")
parser.add_argument("--ent-coef", type=float, default=0.0001, help="Entropy bonus coefficient (applied on resume too)")
parser.add_argument(
"--reset-std",
type=float,
default=None,
help="On resume, reset the policy action std to this value (recovers exploration after entropy collapse)",
)
parser.add_argument("--checkpoint-every", type=int, default=100_000, help="Timesteps between checkpoints")
parser.add_argument("--viz", action="store_true", help="Show game windows (debugging; slow)")
parser.add_argument("--wandb", action="store_true", help="Also log to Weights & Biases")
@@ -67,14 +74,22 @@ def main():
env = VecMonitor(env)
if args.resume:
model = PPO.load(args.resume, env=env, tensorboard_log=str(log_dir))
print(f"Resumed from {args.resume} at {model.num_timesteps} timesteps")
model = PPO.load(args.resume, env=env, tensorboard_log=str(log_dir), ent_coef=args.ent_coef)
print(f"Resumed from {args.resume} at {model.num_timesteps} timesteps (ent_coef={args.ent_coef})")
if args.reset_std is not None:
import math
import torch
with torch.no_grad():
model.policy.log_std.fill_(math.log(args.reset_std))
print(f"Reset policy action std to {args.reset_std}")
else:
model = PPO(
"MultiInputPolicy",
env,
verbose=1,
ent_coef=0.0001,
ent_coef=args.ent_coef,
n_steps=256,
batch_size=256,
learning_rate=3e-4,