added huggingface vocabulary builder

deepchem/feat/vocabulary_builders/hf_vocab.py

@@ -0,0 +1,33 @@
+from deepchem.feat.vocabulary_builders.vocabulary_builder import VocabularyBuilder
+from tokenizers import Tokenizer
+from typing import List
+
+
+class HuggingFaceVocabularyBuilder(VocabularyBuilder):
+
+    def __init__(self, model, trainer):
+        # This tokenizer is from transformers
+        self.model = model
+        self.trainer = trainer
+        self.tokenizer = Tokenizer(model)
+
+    # superclass accepts a DeepChem dataset while huggingface vocabulary builders
+    # reads data from file
+    def build(self, paths: List[str]):  # type: ignore
+
+        self.tokenizer.train(paths, self.trainer)
+
+    @classmethod
+    def load(cls, fname: str):
+        from transformers import PreTrainedTokenizerFast
+        tokenizer = PreTrainedTokenizerFast(tokenizer_file=fname)
+        return tokenizer
+
+    def save(self, fname: str):
+        """
+        Parameters
+        ----------
+        fname: str
+            A json file path
+        """
+        self.tokenizer.save(fname)