{ "selected_gates": 4102.0, "select_seconds": 0.0, "train_seconds": 458.98727710300045, "steps_completed": 240.0, "loss_first": 1.0740166902542114, "loss_last": 1.7048120498657227, "task_loss_first": 1.0740166902542114, "task_loss_last": 1.7048120498657227, "hook_site": "layer_output", "stopped_early": 0.0, "retain_weight": 0.0, "kl_to_base": 0.0, "validation_best_nll": 1.5052597616560872, "validation_best_step": 240.0, "stage": "stage4_2.65b", "checkpoint": "/workspace/RiemannModel/checkpoints/stage4_2.65b", "train_examples": 19000, "validation_examples": 1000, "steps": 240, "lr": 0.005, "gates_requested": 8192, "fit_seconds": 458.9905676841736, "validation_nll_base": 1.5118028684729297, "validation_nll_final": 1.5052597616560872, "warm_start": { "prior_gates": 4631, "carried_gates": 4102, "new_gates_for_grown_dims": 0, "total_gates": 4102, "prior_layers": 36, "prior_hidden": 1600 } }