{ "sft_ckpt": "/ephemeral/ckpts/sft.pt", "pref_path": "/ephemeral/data/preferences.jsonl", "out_ckpt": "/ephemeral/ckpts/reward.pt", "batch_size": 8, "epochs": 1, "eval_steps": 200, "warmup_steps": 50, "lr": 1e-05, "weight_decay": 0.0, "grad_clip": 1.0, "max_len": 768, "save_every": 500 }