{ "selected_gates": 4096.0, "select_seconds": 0.0, "train_seconds": 127.37424483099994, "steps_completed": 240.0, "loss_first": 1.602540373802185, "loss_last": 2.130554437637329, "task_loss_first": 1.602540373802185, "task_loss_last": 2.130554437637329, "hook_site": "layer_output", "stopped_early": 0.0, "retain_weight": 0.0, "kl_to_base": 0.0, "validation_best_nll": 1.9539643727595735, "validation_best_step": 240.0, "stage": "stage2_538m", "checkpoint": "/workspace/RiemannModel/checkpoints/stage2_538m", "train_examples": 19000, "validation_examples": 1000, "steps": 240, "lr": 0.005, "gates_requested": 4096, "fit_seconds": 127.37734150886536, "validation_nll_base": 1.9681743749192033, "validation_nll_final": 1.9539643727595735, "warm_start": { "prior_gates": 4096, "carried_gates": 4096, "new_gates_for_grown_dims": 0, "total_gates": 4096, "prior_layers": 18, "prior_hidden": 1024 } }