Module neuralops.tests.analysis.test_metrics

Tests for neuralops.analysis.metrics module.

Classes

class TestMetrics

Expand source code

class TestMetrics:
    """Test metrics computation and result aggregation."""

    def test_compute_metrics_classification(self):
        # 4 samples, 2 classes
        preds = np.array([[0.8, 0.2], [0.1, 0.9], [0.7, 0.3], [0.3, 0.7]])
        labels = np.array([0, 1, 0, 1])
        res = compute_metrics(preds, labels, task="classification")
        assert "accuracy" in res
        assert res["accuracy"] == 1.0
        assert "f1_macro" in res
        assert "matthews_corrcoef" in res

    def test_compute_metrics_regression(self):
        preds = np.array([1.0, 2.0, 3.0, 4.0])
        labels = np.array([1.1, 1.9, 3.2, 3.9])
        res = compute_metrics(preds, labels, task="regression")
        assert "pearson" in res
        assert "spearman" in res
        assert res["pearson"] > 0.9

    def test_aggregate_results(self):
        records = [
            {"accuracy": 0.80, "loss": 0.5},
            {"accuracy": 0.90, "loss": 0.3},
        ]
        agg = aggregate_results(records)
        assert agg["accuracy_mean"] == pytest.approx(0.85)
        assert agg["loss_mean"] == pytest.approx(0.4)
        assert len(agg["accuracy_values"]) == 2

Test metrics computation and result aggregation.

Methods

def test_aggregate_results(self)

Expand source code

def test_aggregate_results(self):
    records = [
        {"accuracy": 0.80, "loss": 0.5},
        {"accuracy": 0.90, "loss": 0.3},
    ]
    agg = aggregate_results(records)
    assert agg["accuracy_mean"] == pytest.approx(0.85)
    assert agg["loss_mean"] == pytest.approx(0.4)
    assert len(agg["accuracy_values"]) == 2

def test_compute_metrics_classification(self)

Expand source code

def test_compute_metrics_classification(self):
    # 4 samples, 2 classes
    preds = np.array([[0.8, 0.2], [0.1, 0.9], [0.7, 0.3], [0.3, 0.7]])
    labels = np.array([0, 1, 0, 1])
    res = compute_metrics(preds, labels, task="classification")
    assert "accuracy" in res
    assert res["accuracy"] == 1.0
    assert "f1_macro" in res
    assert "matthews_corrcoef" in res

def test_compute_metrics_regression(self)

Expand source code

def test_compute_metrics_regression(self):
    preds = np.array([1.0, 2.0, 3.0, 4.0])
    labels = np.array([1.1, 1.9, 3.2, 3.9])
    res = compute_metrics(preds, labels, task="regression")
    assert "pearson" in res
    assert "spearman" in res
    assert res["pearson"] > 0.9

class TestMetricsTracker

Expand source code

class TestMetricsTracker:
    """Test step metric logging and serialization via JSON lines."""

    def test_tracker_logging_and_flush(self, tmp_path):
        log_file = tmp_path / "metrics.jsonl"
        tracker = MetricsTracker(str(log_file))

        tracker.log({"loss": 0.5, "acc": 0.8}, step=1)
        tracker.log({"loss": 0.3, "acc": 0.9}, step=2, flush=True)

        assert log_file.exists()
        df = tracker.to_dataframe()
        assert len(df) == 2
        assert list(df["step"]) == [1, 2]

Test step metric logging and serialization via JSON lines.

Methods

def test_tracker_logging_and_flush(self, tmp_path)

Expand source code

def test_tracker_logging_and_flush(self, tmp_path):
    log_file = tmp_path / "metrics.jsonl"
    tracker = MetricsTracker(str(log_file))

    tracker.log({"loss": 0.5, "acc": 0.8}, step=1)
    tracker.log({"loss": 0.3, "acc": 0.9}, step=2, flush=True)

    assert log_file.exists()
    df = tracker.to_dataframe()
    assert len(df) == 2
    assert list(df["step"]) == [1, 2]