supermy

/

couplet-helsinki

like 0

Hi supermy,

I noted that you are using opus-mt model with a fast tokenizer and could you please provide some information on that. The Original opus-mt model only support slow tokenizer containing multiple files and the fast tokenizer only need 'tokenizer.json'.

您好，请问下您是如何将opus-mt的tokenizer转换成fast tokenizer，使用一个tokenizer.json文件。非常想知道如何操作，如能告知非常感谢！