Fix bug in pretrained vectorizer sparse embedding formation

GlobalMaksimum / sadedegel

A General Purpose NLP library for Turkish

http://sadedegel.ai

MIT License

93 stars 15 forks source link

Closed dafajon closed 2 years ago

dafajon commented 2 years ago

Change np.float16 -> np.float32. It caused issues in converting scipy.sparse.csr_matrix to numpy.ndarray object.
Implement tests for sparse to dense conversion and model fitting with sklearn API.