Open BoyuanJackChen opened 1 year ago
I realize that there isn't an implementation of stop words in the code yet, so the generation will go all the way till it reaches max length. Any idea how to add stop words in the generation process without fine-tuning the pretrained models?
I realize that there isn't an implementation of stop words in the code yet, so the generation will go all the way till it reaches max length. Any idea how to add stop words in the generation process without fine-tuning the pretrained models?