| {"train/loss": 2.1878, "train/learning_rate": 0.0, "train/epoch": 3.0, "train/global_step": 750, "_timestamp": 1709031404.661967, "_runtime": 3319.244045972824, "_step": 40, "eval/loss": 3.597167491912842, "eval/runtime": 53.6724, "eval/samples_per_second": 37.263, "eval/steps_per_second": 4.658, "train/train_runtime": 2703.4518, "train/train_samples_per_second": 8.878, "train/train_steps_per_second": 0.277, "train/total_flos": 5537821255532544.0, "train/train_loss": 2.995871388753255} |