Skip to content

Commit 8d6a79c

Browse files
committed
Sync CLI defaults with dataclass (100k prompts, 16 rollouts, 32 batch)
1 parent 5161aaf commit 8d6a79c

File tree

1 file changed

+3
-3
lines changed

1 file changed

+3
-3
lines changed

scripts/train_grpo.py

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -488,9 +488,9 @@ def main() -> int:
488488
epilog=__doc__,
489489
)
490490
parser.add_argument("--model", default=None, help="Model name or path")
491-
parser.add_argument("--prompts", type=int, default=10000, help="Number of prompts")
492-
parser.add_argument("--rollouts", type=int, default=8, help="Rollouts per example")
493-
parser.add_argument("--batch-size", type=int, default=16, help="Batch size")
491+
parser.add_argument("--prompts", type=int, default=100000, help="Number of prompts")
492+
parser.add_argument("--rollouts", type=int, default=16, help="Rollouts per example")
493+
parser.add_argument("--batch-size", type=int, default=32, help="Batch size")
494494
parser.add_argument("--lr", type=float, default=1e-6, help="Learning rate")
495495
parser.add_argument("--output", default="models/abide_grpo", help="Output directory")
496496
parser.add_argument("--save-steps", type=int, default=100, help="Save every N steps")

0 commit comments

Comments
 (0)