fix checkpoint naming; add num_epoch parameter (#6277)

This commit is contained in:
YeAnbang
2025-04-26 14:00:28 +08:00
committed by GitHub
parent 26d859f68e
commit 38008858e4
2 changed files with 3 additions and 2 deletions

View File

@@ -130,7 +130,7 @@ class BaseConsumer:
if (step + 1) % self.save_interval == 0 or (step + 1) == self.num_update_per_episode:
if self.rank == 0:
print(f"Start saving policy model at step {step + 1}.")
save_path = os.path.join(self.save_dir, f"modeling-step-{step + 1}")
save_path = os.path.join(self.save_dir, f"modeling-episode-{episode}-step-{step + 1}")
self.booster.save_model(self.policy_model, save_path, shard=True)
if self.rank == 0:
print(f"Saved model checkpoint at step {step + 1} in folder {save_path}")