First, I load the tokenizer using the original file from hf
model_path_baichuan = '../Baichuan2-7B-Base'tokenizer_baichuan = AutoTokenizer.from_pretrained(model_path_baichuan, trust_remote_code=True)
When I just save the tokenizer using
tokenizer_baichuan.save_pretrained(output_folder)
and load it again from the output_foldertokenizer = AutoTokenizer.from_pretrained(output_folder, trust_remote_code=True)
First, I load the tokenizer using the original file from hf
model_path_baichuan = '../Baichuan2-7B-Base'
tokenizer_baichuan = AutoTokenizer.from_pretrained(model_path_baichuan, trust_remote_code=True)
When I just save the tokenizer usingtokenizer_baichuan.save_pretrained(output_folder)
and load it again from theoutput_folder
tokenizer = AutoTokenizer.from_pretrained(output_folder, trust_remote_code=True)
Issue comes: ModuleNotFoundError Traceback (most recent call last)
----> tokenizer = AutoTokenizer.from_pretrained(output_folder, trust_remote_code=True)
ModuleNotFoundError: No module named 'transformers_modules.baichuan_0