{
    "input_size": 7,
    "output_size": 1,
    "hidden_size": 512,
    "n_stacks_trend": 3,
    "n_stacks_seasonal": 4,
    "blocks_per_stack": 3,
    "theta_size_trend": 2,
    "theta_size_season": 4,
    "epochs": 300,
    "batch_size": 32,
    "learning_rate": 0.0001,
    "loss_type": "mse",
    "alpha": 0.1,
    "neuron_function": "GELU(approximate='none')",
    "device": "cuda",
    "train_samples": 1971,
    "validation_samples": 494,
    "train_input_shape": [
        1971,
        7
    ],
    "train_target_shape": [
        1971,
        1
    ],
    "validation_input_shape": [
        494,
        7
    ],
    "validation_target_shape": [
        494,
        1
    ],
    "best_epoch": 300,
    "best_validation_loss": 4.962215825798921e-05,
    "time_complexity": {
        "forward": "O(N * B * H^2)",
        "training_step_approx": "O(N * B * H^2)",
        "gradient_norm_calculation": "O(P)",
        "N_batch_size": 32,
        "B_total_blocks": 21,
        "H_hidden_size": 512,
        "I_input_size": 7,
        "O_output_size": 1,
        "total_parameters": 16750308
    },
    "big_o_empirical_test": {
        "benchmark_batch_size": 8,
        "device": "cuda",
        "hidden_size_scaling": {
            "expected_complexity": "O(H^2)",
            "results": [
                {
                    "hidden_size": 128,
                    "total_blocks": 21,
                    "batch_size": 8,
                    "time_seconds_median": 0.01987130008637905,
                    "time_milliseconds_median": 19.87130008637905,
                    "theoretical_relative_cost_vs_first": 1.0
                },
                {
                    "hidden_size": 256,
                    "total_blocks": 21,
                    "batch_size": 8,
                    "time_seconds_median": 0.01879219990223646,
                    "time_milliseconds_median": 18.79219990223646,
                    "theoretical_relative_cost_vs_first": 4.0
                },
                {
                    "hidden_size": 512,
                    "total_blocks": 21,
                    "batch_size": 8,
                    "time_seconds_median": 0.019390400033444166,
                    "time_milliseconds_median": 19.390400033444166,
                    "theoretical_relative_cost_vs_first": 16.0
                }
            ],
            "log_log_fit": {
                "exponent": -0.017671848609463185,
                "intercept": -3.847262919698698,
                "r2": 0.1915588923022309
            }
        },
        "block_scaling": {
            "expected_complexity": "O(B)",
            "results": [
                {
                    "blocks_per_stack": 10,
                    "total_blocks": 70,
                    "hidden_size": 512,
                    "batch_size": 8,
                    "time_seconds_median": 0.06300389999523759,
                    "time_milliseconds_median": 63.00389999523759,
                    "theoretical_relative_cost_vs_first": 1.0
                },
                {
                    "blocks_per_stack": 20,
                    "total_blocks": 140,
                    "hidden_size": 512,
                    "batch_size": 8,
                    "time_seconds_median": 0.12464780011214316,
                    "time_milliseconds_median": 124.64780011214316,
                    "theoretical_relative_cost_vs_first": 2.0
                },
                {
                    "blocks_per_stack": 30,
                    "total_blocks": 210,
                    "hidden_size": 512,
                    "batch_size": 8,
                    "time_seconds_median": 0.18604329996742308,
                    "time_milliseconds_median": 186.04329996742308,
                    "theoretical_relative_cost_vs_first": 3.0
                }
            ],
            "log_log_fit": {
                "exponent": 0.9854571097543345,
                "intercept": -6.951476412790202,
                "r2": 0.9999991876654754
            }
        },
        "input_batch_scaling": {
            "expected_complexity": "O(N)",
            "results": [
                {
                    "batch_size": 1,
                    "hidden_size": 512,
                    "total_blocks": 21,
                    "time_seconds_median": 0.01942260004580021,
                    "time_milliseconds_median": 19.42260004580021,
                    "theoretical_relative_cost_vs_first": 1.0
                },
                {
                    "batch_size": 2,
                    "hidden_size": 512,
                    "total_blocks": 21,
                    "time_seconds_median": 0.019125699996948242,
                    "time_milliseconds_median": 19.125699996948242,
                    "theoretical_relative_cost_vs_first": 2.0
                },
                {
                    "batch_size": 4,
                    "hidden_size": 512,
                    "total_blocks": 21,
                    "time_seconds_median": 0.018762200139462948,
                    "time_milliseconds_median": 18.762200139462948,
                    "theoretical_relative_cost_vs_first": 4.0
                },
                {
                    "batch_size": 8,
                    "hidden_size": 512,
                    "total_blocks": 21,
                    "time_seconds_median": 0.01835409994237125,
                    "time_milliseconds_median": 18.35409994237125,
                    "theoretical_relative_cost_vs_first": 8.0
                },
                {
                    "batch_size": 16,
                    "hidden_size": 512,
                    "total_blocks": 21,
                    "time_seconds_median": 0.019410799955949187,
                    "time_milliseconds_median": 19.410799955949187,
                    "theoretical_relative_cost_vs_first": 16.0
                },
                {
                    "batch_size": 32,
                    "hidden_size": 512,
                    "total_blocks": 21,
                    "time_seconds_median": 0.01993660000152886,
                    "time_milliseconds_median": 19.93660000152886,
                    "theoretical_relative_cost_vs_first": 32.0
                }
            ],
            "log_log_fit": {
                "exponent": 0.006306562161759777,
                "intercept": -3.965757991421921,
                "r2": 0.07918679416134755
            }
        }
    }
}