stats.json

{"world_size": 1, "epochs": 2, "steps": 4371, "seqs": 9550, "tokens": 7050446, "last_epoch_steps": 0, "last_epoch_seqs": 0, "last_epoch_tokens": 0, "experiment_tracking_run_id": null, "loss_ema": 0.7818833695804316, "loss_sum": 30.961968302726746, "mtp_loss_ema": 0, "mtp_loss_sum": 0, "eval_losses_avg": [0.9667969819826957, 0.8918106800470597, 0.844005211805686, 0.8168310828697987, 0.7929086157908807, 0.79764097623336, 0.7486214874646603, 0.7394010134232349, 0.7235642022047287, 0.7187379384652163, 0.7143815450179272, 0.7060007116733453]}