first_layer_2/model_6000/training_state.json

{
    "global_step": 6000,
    "update_step": 6000,
    "tokens_seen": 3833905152,
    "tokens_seen_before": 3833266176,
    "update_time": 2.875896453857422,
    "wandb_id": "f3ljzhyw"
}