YifanXu74 6fc1e46e79 init
2023-10-07 23:02:26 +08:00

19 lines
567 B
Python

from .simple_tokenizer import SimpleTokenizer
def build_tokenizer(tokenizer_name):
tokenizer = None
if tokenizer_name == 'clip':
tokenizer = SimpleTokenizer()
elif 'hf_' in tokenizer_name:
from .hfpt_tokenizer import HFPTTokenizer
tokenizer = HFPTTokenizer(pt_name=tokenizer_name[3:])
elif 'hfc_' in tokenizer_name:
from .hfpt_tokenizer import HFPTTokenizer
tokenizer = HFPTTokenizer(pt_name=tokenizer_name[4:])
else:
raise ValueError('Unknown tokenizer')
return tokenizer