| [ | |
| { | |
| "loss": 3.3461, | |
| "grad_norm": 0.810087263584137, | |
| "learning_rate": 1.8e-05, | |
| "entropy": 2.415625, | |
| "num_tokens": 1129.0, | |
| "mean_token_accuracy": 0.408928906917572, | |
| "epoch": 0.8333333333333334, | |
| "step": 10 | |
| }, | |
| { | |
| "train_runtime": 295.1505, | |
| "train_samples_per_second": 0.041, | |
| "train_steps_per_second": 0.041, | |
| "total_flos": 876711458304.0, | |
| "train_loss": 3.3814881642659507, | |
| "entropy": 2.4609375, | |
| "num_tokens": 1364.0, | |
| "mean_token_accuracy": 0.39988937973976135, | |
| "epoch": 1.0, | |
| "step": 12 | |
| } | |
| ] |