Fix cross-validation class counts (#32)

Author: henry3260

llm_fingerprinter/classifier.py

@@ -610,8 +610,8 @@ def cross_validate(self, X: np.ndarray, y: np.ndarray, n_folds: int = 5):
         from sklearn.model_selection import StratifiedKFold
         from sklearn.metrics import precision_recall_fscore_support, confusion_matrix
 
-        unique_classes = np.unique(y)
-        actual_folds = min(n_folds, min(np.bincount(y.astype(int))))
+        _, class_counts = np.unique(y, return_counts=True)
+        actual_folds = min(n_folds, int(class_counts.min()))
         if actual_folds < 2:
             logger.warning("Not enough samples per class for cross-validation")
             return None

tests/classifier/test_classifier_cross_validation.py

@@ -0,0 +1,33 @@
+import numpy as np
+
+from llm_fingerprinter.classifier import EnsembleClassifier
+
+
+def test_cross_validate_ignores_configured_classes_without_samples(monkeypatch):
+    classifier = EnsembleClassifier(
+        model_families={"alpha": 0, "missing": 1, "beta": 2},
+        augment_data=False,
+    )
+    X = np.array(
+        [[0.0], [0.1], [0.2], [1.0], [1.1], [1.2]],
+        dtype=np.float32,
+    )
+    y = np.array([0, 0, 0, 2, 2, 2])
+
+    monkeypatch.setattr(EnsembleClassifier, "train", lambda self, X, y: True)
+    monkeypatch.setattr(
+        EnsembleClassifier,
+        "predict_with_confidence",
+        lambda self, fingerprint: (
+            "alpha" if fingerprint[0] < 0.5 else "beta",
+            1.0,
+            {},
+            {},
+        ),
+    )
+
+    results = classifier.cross_validate(X, y, n_folds=5)
+
+    assert results is not None
+    assert results["n_folds"] == 3
+    assert results["mean_accuracy"] == 1.0