@dataclass
class codebert_output:
embedding: np.ndarray
probability: float
codeBERT's probability is not passed down to the MLP classifier, only the embedding is passed down.
(Less of an issue, more of something to be that can be changed and re-run the pipeline to see if this affects accuracy and how it affects how the classifier learns eg does it become dependent on codeBERT's probability output and ignore embeddings altogether)
codeBERT's probability is not passed down to the MLP classifier, only the embedding is passed down.
(Less of an issue, more of something to be that can be changed and re-run the pipeline to see if this affects accuracy and how it affects how the classifier learns eg does it become dependent on codeBERT's probability output and ignore embeddings altogether)