first_layer_2/model_8000/training_state.json

{
    "global_step": 8000,
    "update_step": 8000,
    "tokens_seen": 5111857152,
    "tokens_seen_before": 5111218176,
    "update_time": 2.87611722946167,
    "wandb_id": "f3ljzhyw"
}