Suburst's picture
Upload 27 files
f8bd4d2 verified
import json
class Tokenizer:
def __init__(self, word2int_path, int2word_path):
with open(word2int_path, 'r') as f:
self.word2int = json.load(f)
with open(int2word_path, 'r') as f:
self.int2word = {int(k): v for k, v in json.load(f).items()}
def encode(self, word):
return self.word2int[word]
def decode(self, int_val):
return self.int2word[int_val]