Merge pull request #2578 from seyonechithrananda/st_fix

import os

import unittest

from unittest import TestCase

import pytest

try:

  from transformers import RobertaForMaskedLM

  from deepchem.feat.smiles_tokenizer import SmilesTokenizer

  """Tests the SmilesTokenizer to load the USPTO vocab file and a ChemBERTa Masked LM model with pre-trained weights.."""

  @unittest.skipIf(not has_transformers, 'transformers are not installed')

  @pytest.mark.torch

  def test_tokenize(self):

    current_dir = os.path.dirname(os.path.realpath(__file__))

    vocab_path = os.path.join(current_dir, 'data', 'vocab.txt')

    tokenized_smiles = [

        12, 16, 16, 16, 17, 16, 16, 18, 16, 19, 16, 17, 22, 19, 18, 33, 17, 16,

        18, 23, 181, 17, 22, 19, 18, 17, 19, 16, 33, 20, 19, 55, 17, 16, 23, 18,

        17, 33, 17, 19, 18, 35, 20, 19, 18, 16, 20, 22, 16, 16, 22, 16, 21, 23,

        20, 23, 22, 16, 23, 22, 16, 21, 23, 18, 19, 16, 20, 22, 16, 16, 22, 16,

        16, 22, 16, 20, 13

        18, 23, 181, 17, 22, 19, 18, 17, 19, 16, 33, 20, 19, 55, 17, 16, 38, 23,

        18, 17, 33, 17, 19, 18, 35, 20, 19, 18, 16, 20, 22, 16, 16, 22, 16, 21,

        23, 20, 23, 22, 16, 23, 22, 16, 21, 23, 18, 19, 16, 20, 22, 16, 16, 22,

        16, 16, 22, 16, 20, 13

    ]

    model = RobertaForMaskedLM.from_pretrained(

  - numpy==1.21.*

  - pip==20.2.*

  - pip:

    - transformers==4.6.*

    - tensorflow~=2.4

    - dgllife==0.2.8

    - pymatgen==2020.12.31