update IQA results

2025-09-09 12:51:55 +07:00
parent eb465155d1
commit eb0a12448f
32 changed files with 3551 additions and 4 deletions
--- a/docs/task/cni/results/facture_confusion_matrix.png
+++ b/docs/task/cni/results/facture_confusion_matrix.png
--- a/docs/task/cni/results/facture_deqa_images.xlsx
+++ b/docs/task/cni/results/facture_deqa_images.xlsx
--- a/docs/task/cni/results/facture_metric_curves.png
+++ b/docs/task/cni/results/facture_metric_curves.png
--- a/docs/task/cni/results/facture_precision_recall_curve.png
+++ b/docs/task/cni/results/facture_precision_recall_curve.png
--- a/docs/task/cni/results/facture_roc_like_curve.png
+++ b/docs/task/cni/results/facture_roc_like_curve.png
--- a/docs/task/cni/results/facture_score_distributions.png
+++ b/docs/task/cni/results/facture_score_distributions.png
--- a/docs/task/cni/results/facture_sorted_scores_with_thr.png
+++ b/docs/task/cni/results/facture_sorted_scores_with_thr.png
--- a/docs/task/facture/200_samples/facture_confusion_matrix.png
+++ b/docs/task/facture/200_samples/facture_confusion_matrix.png
--- a/docs/task/facture/200_samples/facture_decisions.csv
+++ b/docs/task/facture/200_samples/facture_decisions.csv
--- a/docs/task/facture/200_samples/facture_deqa_images.xlsx
+++ b/docs/task/facture/200_samples/facture_deqa_images.xlsx
--- a/docs/task/facture/200_samples/facture_metric_curves.png
+++ b/docs/task/facture/200_samples/facture_metric_curves.png
--- a/docs/task/facture/200_samples/facture_precision_recall_curve.png
+++ b/docs/task/facture/200_samples/facture_precision_recall_curve.png
--- a/docs/task/facture/200_samples/facture_roc_like_curve.png
+++ b/docs/task/facture/200_samples/facture_roc_like_curve.png
--- a/docs/task/facture/200_samples/facture_score_distributions.png
+++ b/docs/task/facture/200_samples/facture_score_distributions.png
--- a/docs/task/facture/200_samples/facture_sorted_scores_with_thr.png
+++ b/docs/task/facture/200_samples/facture_sorted_scores_with_thr.png
--- a/docs/task/facture/200_samples/facture_thresholds_summary.json
+++ b/docs/task/facture/200_samples/facture_thresholds_summary.json
--- a/docs/task/facture/full_samples/facture_confusion_matrix.png
+++ b/docs/task/facture/full_samples/facture_confusion_matrix.png
--- a/docs/task/facture/full_samples/facture_decisions.csv
+++ b/docs/task/facture/full_samples/facture_decisions.csv
--- a/docs/task/facture/full_samples/facture_deqa_images.xlsx
+++ b/docs/task/facture/full_samples/facture_deqa_images.xlsx
--- a/docs/task/facture/full_samples/facture_metric_curves.png
+++ b/docs/task/facture/full_samples/facture_metric_curves.png
--- a/docs/task/facture/full_samples/facture_precision_recall_curve.png
+++ b/docs/task/facture/full_samples/facture_precision_recall_curve.png
--- a/docs/task/facture/full_samples/facture_roc_like_curve.png
+++ b/docs/task/facture/full_samples/facture_roc_like_curve.png
--- a/docs/task/facture/full_samples/facture_score_distributions.png
+++ b/docs/task/facture/full_samples/facture_score_distributions.png
--- a/docs/task/facture/full_samples/facture_sorted_scores_with_thr.png
+++ b/docs/task/facture/full_samples/facture_sorted_scores_with_thr.png
--- a/docs/task/facture/full_samples/facture_thresholds_summary.json
+++ b/docs/task/facture/full_samples/facture_thresholds_summary.json
@@ -0,0 +1,50 @@
+{
+  "positive_definition": "HIGH when score >= threshold",
+  "best_thresholds": {
+    "f1": {
+      "threshold": 1.6,
+      "value": 0.9249762583095917,
+      "confusion": {
+        "TP": 2922,
+        "FP": 474,
+        "FN": 0,
+        "TN": 0
+      }
+    },
+    "accuracy": {
+      "threshold": 1.6,
+      "value": 0.8604240282685512,
+      "confusion": {
+        "TP": 2922,
+        "FP": 474,
+        "FN": 0,
+        "TN": 0
+      }
+    },
+    "precision": {
+      "threshold": 4.2,
+      "value": 1.0,
+      "confusion": {
+        "TP": 3,
+        "FP": 0,
+        "FN": 2919,
+        "TN": 474
+      }
+    },
+    "recall": {
+      "threshold": 1.6,
+      "value": 1.0,
+      "confusion": {
+        "TP": 2922,
+        "FP": 474,
+        "FN": 0,
+        "TN": 0
+      }
+    }
+  },
+  "counts": {
+    "total": 3396,
+    "positives": 2922,
+    "negatives": 474
+  }
+}
--- a/docs/task/facture/results/facture_confusion_matrix.png
+++ b/docs/task/facture/results/facture_confusion_matrix.png
--- a/docs/task/facture/results/facture_metric_curves.png
+++ b/docs/task/facture/results/facture_metric_curves.png
--- a/docs/task/facture/results/facture_precision_recall_curve.png
+++ b/docs/task/facture/results/facture_precision_recall_curve.png
--- a/docs/task/facture/results/facture_roc_like_curve.png
+++ b/docs/task/facture/results/facture_roc_like_curve.png
--- a/docs/task/facture/results/facture_score_distributions.png
+++ b/docs/task/facture/results/facture_score_distributions.png
--- a/docs/task/facture/results/facture_sorted_scores_with_thr.png
+++ b/docs/task/facture/results/facture_sorted_scores_with_thr.png