{ "epoch": 2.0, "eval_accuracy": 0.9172453703703703, "eval_loss": 0.418969064950943, "eval_margin": 0.8687698929398148, "eval_max_reward": 1.382568359375, "eval_mean_reward": -0.11189778645833333, "eval_min_reward": -1.5139612268518519, "eval_num_tokens": 92427959.0, "eval_runtime": 403.7554, "eval_samples_per_second": 4.275, "eval_steps_per_second": 0.134 }