Merge pull request #50 from robin-janssen/improve-surrogate-evaluation

Improve surrogate evaluation

Author: robin-janssen

codes/benchmark/__init__.py

@@ -1,19 +1,19 @@
 from .bench_fcts import (
     compare_batchsize,
-    compare_dynamic_accuracy,
+    compare_errors,
     compare_extrapolation,
+    compare_gradients,
     compare_inference_time,
     compare_interpolation,
     compare_main_losses,
     compare_models,
-    compare_relative_errors,
     compare_sparse,
     compare_UQ,
     evaluate_accuracy,
     evaluate_batchsize,
     evaluate_compute,
-    evaluate_dynamic_accuracy,
     evaluate_extrapolation,
+    evaluate_gradients,
     evaluate_interpolation,
     evaluate_sparse,
     evaluate_UQ,
@@ -24,15 +24,12 @@
 from .bench_plots import (
     get_custom_palette,
     inference_time_bar_plot,
-    int_ext_sparse,
     plot_all_generalization_errors,
     plot_average_errors_over_time,
     plot_average_uncertainty_over_time,
-    plot_comparative_dynamic_correlation_heatmaps,
     plot_comparative_error_correlation_heatmaps,
+    plot_comparative_gradient_heatmaps,
     plot_dynamic_correlation,
-    plot_dynamic_correlation_heatmap,
-    plot_error_correlation_heatmap,
     plot_error_distribution_comparative,
     plot_error_distribution_per_quantity,
     plot_error_percentiles_over_time,
@@ -41,13 +38,15 @@
     plot_example_predictions_with_uncertainty,
     plot_generalization_error_comparison,
     plot_generalization_errors,
+    plot_gradients_heatmap,
     plot_loss_comparison,
     plot_loss_comparison_train_duration,
     plot_losses,
     plot_MAE_comparison,
     plot_relative_errors,
     plot_surr_losses,
     plot_uncertainty_confidence,
+    plot_uncertainty_heatmap,
     plot_uncertainty_over_time_comparison,
     plot_uncertainty_vs_errors,
     rel_errors_and_uq,
@@ -65,6 +64,7 @@
     flatten_dict,
     format_seconds,
     format_time,
+    format_value,
     get_model_config,
     get_required_models_list,
     get_surrogate,
@@ -80,7 +80,7 @@
 __all__ = [
     "run_benchmark",
     "evaluate_accuracy",
-    "evaluate_dynamic_accuracy",
+    "evaluate_gradients",
     "time_inference",
     "evaluate_compute",
     "evaluate_interpolation",
@@ -90,9 +90,9 @@
     "evaluate_UQ",
     "compare_models",
     "compare_main_losses",
-    "compare_relative_errors",
+    "compare_errors",
     "compare_inference_time",
-    "compare_dynamic_accuracy",
+    "compare_gradients",
     "compare_interpolation",
     "compare_extrapolation",
     "compare_sparse",
@@ -120,14 +120,13 @@
     "plot_uncertainty_over_time_comparison",
     "inference_time_bar_plot",
     "plot_generalization_error_comparison",
-    "plot_error_correlation_heatmap",
-    "plot_dynamic_correlation_heatmap",
+    "plot_uncertainty_heatmap",
+    "plot_gradients_heatmap",
     "plot_error_distribution_comparative",
     "plot_error_percentiles_over_time",
     "plot_comparative_error_correlation_heatmaps",
-    "plot_comparative_dynamic_correlation_heatmaps",
+    "plot_comparative_gradient_heatmaps",
     "get_custom_palette",
-    "int_ext_sparse",
     "plot_all_generalization_errors",
     "rel_errors_and_uq",
     "check_surrogate",
@@ -144,6 +143,7 @@
     "get_surrogate",
     "format_time",
     "format_seconds",
+    "format_value",
     "flatten_dict",
     "convert_dict_to_scientific_notation",
     "make_comparison_csv",