When I encountered the error below, I solved the problem by rewriting the tokenizer as follows.
/uth-bert/tokenization_mod.py in load_vocab(vocab_file)
122 vocab = collections.OrderedDict()
123 index = 0
--> 124 with tf.gfile.GFile(vocab_file, "r") as reader:
125 while True:
126 token = convert_to_unicode(reader.readline())
AttributeError: module 'tensorflow' has no attribute 'gfile'
→→replace tf.gfile.GFile to tf.io.gfile.GFile
If UTH-BERT intends to support tf2 series, I think it is better to modify this part.
When I encountered the error below, I solved the problem by rewriting the tokenizer as follows.
→→replace tf.gfile.GFile to tf.io.gfile.GFile
If UTH-BERT intends to support tf2 series, I think it is better to modify this part.