TensorMachinesRegularizedLeastSquares expects a d3m.container.numpy.ndarray as input
One of the D3M goals/requirements is that primitives of a certain type (regression for example) need to be swappable in a pipeline. d3m.primitives.regression.tensor_machines_regularized_least_squares.TensorMachinesRegularizedLeastSquares breaks this rule and so is not considered by our system for TA2 pipeline search.
runtime = Runtime(pipeline=pipeline, problem_description=problem_description, context=metadata_base.Context.TESTING)
File "/ta3ta2-api/src/d3m/d3m/runtime.py", line 220, in __init__
self.pipeline.check(allow_placeholders=False, standard_pipeline=self.is_standard_pipeline)
File "/ta3ta2-api/src/d3m/d3m/metadata/pipeline.py", line 1482, in check
self._check(allow_placeholders, standard_pipeline, input_types)
File "/ta3ta2-api/src/d3m/d3m/metadata/pipeline.py", line 1622, in _check
input_type=type_info.structural_type,
d3m.exceptions.InvalidPipelineError: Argument 'inputs' of step 8 of pipeline 'ebe661da-8d95-4c76-9a6b-a557a1df9d2c' has type '<class 'd3m.container.numpy.ndarray'>', but it is getting a type '<class 'd3m.container.pandas.DataFrame'>'.
TensorMachinesRegularizedLeastSquares is looking for a d3m.container.numpy.ndarray but a d3m.container.pandas.DataFrame is expected to be acceptable.