{ "selected_gates": 6732.0, "select_seconds": 0.0, "train_seconds": 2026.6077014809998, "steps_completed": 240.0, "loss_first": 0.9671240448951721, "loss_last": 1.4579845666885376, "task_loss_first": 0.9671240448951721, "task_loss_last": 1.4579845666885376, "hook_site": "layer_output", "stopped_early": 0.0, "retain_weight": 0.0, "kl_to_base": 0.0, "validation_best_nll": 1.3103517218016936, "validation_best_step": 240.0, "stage": "stage6_12.85b", "checkpoint": "/workspace/RiemannModel/checkpoints/stage6_12.85b", "train_examples": 19000, "validation_examples": 1000, "steps": 240, "lr": 0.005, "gates_requested": 8192, "fit_seconds": 2026.6114346981049, "validation_nll_base": 1.3227896775331283, "validation_nll_final": 1.3103517218016936, "warm_start": { "prior_gates": 4102, "carried_gates": 4102, "new_gates_for_grown_dims": 2630, "total_gates": 6732, "prior_layers": 32, "prior_hidden": 4096 } }