Units Tests Failing

[raabrp]

In order to get tests to pass on a fresh clone of the repository, I had to make a few edits and set a timeout for pytest. After setting the timeout, some tests continue to fail.

Edits I had to make

• test_loss.py (replace all occurances of np.float -> float)
• test_blocks.py (import KANBlock and StackedMLP from neuromancer.modules.blocks)

Unit Tests Still Failing

FAILED tests/psl/test_norm.py::test_emulator_integration4[LotkaVolterra] - Failed: Timeout >10.0s
FAILED tests/test_blocks.py::test_linear_shape - Failed: Timeout >10.0s
FAILED tests/test_blocks.py::test_res_mlp_shape - hypothesis.errors.FlakyStrategyDefinition: Inconsistent data generation! Data generation behaved differently be...
FAILED tests/test_blocks.py::test_bilinear_shape - hypothesis.errors.FlakyFailure: Hypothesis test_bilinear_shape(batchsize=426, insize=426, outsize=258, lin=Spec...
FAILED tests/test_constraints.py::test_constraint_device_placement - AssertionError: Torch not compiled with CUDA enabled
FAILED tests/test_trainer.py::test_train_runs_for_epochs[get_problem0-data_setup_function-10] - TypeError: LitTrainer.fit() missing 1 required positional argument: 'data_setup_function'
FAILED tests/test_trainer.py::test_weight_updates[get_problem0-data_setup_function] - TypeError: LitTrainer.fit() missing 1 required positional argument: 'data_setup_function'
FAILED tests/test_utils.py::test_handle_device_placement[callable_1] - AssertionError: Torch not compiled with CUDA enabled
FAILED tests/test_utils.py::test_handle_device_placement[callable_2] - AssertionError: Torch not compiled with CUDA enabled
FAILED tests/test_utils.py::test_handle_device_placement[callable_3] - AssertionError: Torch not compiled with CUDA enabled

Additional Failure after commenting the above (Non-deterministic?)

test_blocks.py: test_basis_linear_shape

raises .venv/lib/python3.10/site-packages/neuromancer/slim/linear.py:683: TypeError

def forward(self, x):
    x = self.Umultiply(x)
    x = torch.matmul(x, self.Sigma())
    x = self.Vmultiply(x)

  return x + self.bias

E TypeError: unsupported operand type(s) for +: 'Tensor' and 'NoneType'
E Falsifying example: test_basis_linear_shape(
E batchsize=1, # or any other generated value
E insize=1, # or any other generated value
E outsize=1, # or any other generated value
E bias=False,
E lin=neuromancer.slim.linear.SpectralLinear,
E )