tests/test_model.py

from __future__ import annotations

import torch

from deepgenopix.model import DeepGenomicClassifier


def test_classifier_pads_to_stride_before_compression():
    model = DeepGenomicClassifier(num_classes=2, compressor_stride=4)
    x = torch.randn(1, 3, 5)
    lengths = torch.tensor([5])

    padded = model._pad_to_stride(x, lengths)
    assert padded.shape == (1, 3, 8)


def test_classifier_uses_ceil_token_lengths_for_masks():
    model = DeepGenomicClassifier(num_classes=2, compressor_stride=4)
    x = torch.zeros(1, 64, 2)
    lengths = torch.tensor([5])

    mask, comp_lens = model._compute_mask(x, lengths)
    assert comp_lens.tolist() == [2]
    assert mask.shape == (1, 2)
    assert mask.tolist() == [[False, False]]