Use torch.testing.assert_close instead of approx

The tests outputs `RuntimeError: Can't call numpy() on Tensor that
requires grad. Use tensor.detach().numpy() instead.` for torch above
v1.4. The error seems to came from trying to convert tensor to scalar.

Replace the pytest approx() with `torch.testing.assert_close()`
to let torch handle the conversion.
Use `assert_allclose()` for torch versions before 1.9.0.

requirements-test.txt

@@ -4,3 +4,4 @@
 torch
 pytest>=8,<9
 pytest-cov>=2,<3
+packaging

tests/test_crf.py

@@ -6,6 +6,7 @@
 import pytest
 import torch
 import torch.nn as nn
+from packaging.version import Version
 
 from torchcrf import CRF
 
@@ -14,6 +15,8 @@
 random.seed(RANDOM_SEED)
 torch.manual_seed(RANDOM_SEED)
 
+HAS_ASSERT_CLOSE = Version(torch.__version__) >= Version("1.9.0")
+
 
 def compute_score(crf, emission, tag):
     # emission: (seq_length, num_tags)
@@ -116,7 +119,10 @@ def test_works_with_mask(self):
             denominator = math.log(sum(math.exp(s) for s in all_scores))
             manual_llh += numerator - denominator
 
-        assert llh.item() == approx(manual_llh)
+        if HAS_ASSERT_CLOSE:
+            torch.testing.assert_close(llh, manual_llh, atol=1e-12, rtol=1e-6)
+        else:
+            torch.testing.assert_allclose(llh, manual_llh, atol=1e-12, rtol=1e-6)
         llh.backward()  # ensure gradients can be computed
 
     def test_works_without_mask(self):
@@ -186,8 +192,10 @@ def test_reduction_none(self):
             denominator = math.log(sum(math.exp(s) for s in all_scores))
             manual_llh.append(numerator - denominator)
 
-        for llh_, manual_llh_ in zip(llh, manual_llh):
-            assert llh_.item() == approx(manual_llh_)
+        if HAS_ASSERT_CLOSE:
+            torch.testing.assert_close(llh, torch.tensor(manual_llh), atol=1e-12, rtol=1e-6)
+        else:
+            torch.testing.assert_allclose(llh, torch.tensor(manual_llh), atol=1e-12, rtol=1e-6)
 
     def test_reduction_mean(self):
         crf = make_crf()
@@ -219,7 +227,10 @@ def test_reduction_mean(self):
             denominator = math.log(sum(math.exp(s) for s in all_scores))
             manual_llh += numerator - denominator
 
-        assert llh.item() == approx(manual_llh / batch_size)
+        if HAS_ASSERT_CLOSE:
+            torch.testing.assert_close(llh, manual_llh / batch_size, atol=1e-12, rtol=1e-6)
+        else:
+            torch.testing.assert_allclose(llh, manual_llh / batch_size, atol=1e-12, rtol=1e-6)
 
     def test_reduction_token_mean(self):
         crf = make_crf()
@@ -258,7 +269,10 @@ def test_reduction_token_mean(self):
             manual_llh += numerator - denominator
             n_tokens += seq_len
 
-        assert llh.item() == approx(manual_llh / n_tokens)
+        if HAS_ASSERT_CLOSE:
+            torch.testing.assert_close(llh, manual_llh / n_tokens, atol=1e-12, rtol=1e-6)
+        else:
+            torch.testing.assert_allclose(llh, manual_llh / n_tokens, atol=1e-12, rtol=1e-6)
 
     def test_batch_first(self):
         crf = make_crf()