qwen3-8b-base-kto-ultrafeedback-4xh200-batch-128/train_results.json

{
    "epoch": 1.0,
    "total_flos": 0.0,
    "train_loss": 1.4430672781629712,
    "train_runtime": 11696.6719,
    "train_samples": 122270,
    "train_samples_per_second": 10.453,
    "train_steps_per_second": 0.082
}