Skip tests when there are import issues on CI worker (meta-pytorch#132)

* move the test that doesn't run on CI to a new folder

* Revert "move the test that doesn't run on CI to a new folder"

This reverts commit b2fd5c3.

* move the right file...

* use pytest skipif

* catch all exceptions

* change function name

* fix

* typo

Author: casteryh

tests/unit_tests/actors/test_reference_actor.py renamed to tests/unit_tests/test_reference_actor.py

@@ -8,17 +8,32 @@
 Tests for reference_actor.py - compute_logprobs function
 """
 
+import unittest
+
 import pytest
 import torch
 
-from forge.actors.reference_actor import compute_logprobs
 
+def _import_error():
+    try:
+        import forge.actors.reference_actor  # noqa: F401
+
+        return False
+    except Exception:
+        return True
 
-class TestComputeLogprobs:
+
+class TestComputeLogprobs(unittest.TestCase):
     """Test the compute_logprobs utility function."""
 
+    @pytest.mark.skipif(
+        _import_error(),
+        reason="Import error, likely due to missing dependencies on CI.",
+    )
     def test_compute_logprobs_basic(self):
         """Test basic logprobs computation."""
+        from forge.actors.reference_actor import compute_logprobs
+
         batch_size = 1
         seq_len = 5
         vocab_size = 1000
@@ -36,8 +51,14 @@ def test_compute_logprobs_basic(self):
         assert result.shape == (batch_size, response_len)
         assert torch.all(result <= 0)  # Log probabilities should be <= 0
 
+    @pytest.mark.skipif(
+        _import_error(),
+        reason="Import error, likely due to missing dependencies on CI.",
+    )
     def test_compute_logprobs_with_temperature(self):
         """Test logprobs computation with temperature scaling."""
+        from forge.actors.reference_actor import compute_logprobs
+
         batch_size = 1
         seq_len = 5
         vocab_size = 1000
@@ -55,8 +76,14 @@ def test_compute_logprobs_with_temperature(self):
         default_result = compute_logprobs(logits, input_ids)
         assert not torch.allclose(result, default_result)
 
+    @pytest.mark.skipif(
+        _import_error(),
+        reason="Import error, likely due to missing dependencies on CI.",
+    )
     def test_compute_logprobs_single_token(self):
         """Test logprobs computation with single token response."""
+        from forge.actors.reference_actor import compute_logprobs
+
         batch_size = 1
         seq_len = 5
         vocab_size = 1000
@@ -70,8 +97,14 @@ def test_compute_logprobs_single_token(self):
         assert result.shape == (batch_size, response_len)
         assert result.numel() == 1  # Single element
 
+    @pytest.mark.skipif(
+        _import_error(),
+        reason="Import error, likely due to missing dependencies on CI.",
+    )
     def test_compute_logprobs_empty_response(self):
         """Test logprobs computation with empty response."""
+        from forge.actors.reference_actor import compute_logprobs
+
         batch_size = 1
         seq_len = 5
         vocab_size = 1000
@@ -84,8 +117,14 @@ def test_compute_logprobs_empty_response(self):
 
         assert result.shape == (batch_size, response_len)
 
+    @pytest.mark.skipif(
+        _import_error(),
+        reason="Import error, likely due to missing dependencies on CI.",
+    )
     def test_compute_logprobs_empty_prompt(self):
         """Test logprobs computation with empty prompt."""
+        from forge.actors.reference_actor import compute_logprobs
+
         batch_size = 1
         vocab_size = 1000
         prompt_len = 0