{ "selected_gates": 4631.0, "select_seconds": 0.0, "train_seconds": 254.6721720530004, "steps_completed": 240.0, "loss_first": 1.200066089630127, "loss_last": 1.8443958759307861, "task_loss_first": 1.200066089630127, "task_loss_last": 1.8443958759307861, "hook_site": "layer_output", "stopped_early": 0.0, "retain_weight": 0.0, "kl_to_base": 0.0, "validation_best_nll": 1.6178833635073087, "validation_best_step": 240.0, "stage": "stage3_1.19b", "checkpoint": "/workspace/RiemannModel/checkpoints/stage3_1.19b", "train_examples": 19000, "validation_examples": 1000, "steps": 240, "lr": 0.005, "gates_requested": 8192, "fit_seconds": 254.67549347877502, "validation_nll_base": 1.6293330653363551, "validation_nll_final": 1.6178833635073087, "warm_start": { "prior_gates": 4096, "carried_gates": 4096, "new_gates_for_grown_dims": 535, "total_gates": 4631, "prior_layers": 24, "prior_hidden": 1280 } }