first_layer_2/model_5000/training_state.json

{
    "global_step": 5000,
    "update_step": 5000,
    "tokens_seen": 3194929152,
    "tokens_seen_before": 3194290176,
    "update_time": 2.8782315254211426,
    "wandb_id": "f3ljzhyw"
}