Text Generation
PyTorch
12 languages
causal-lm
rwkv

ChatML as prompt template

#3
by vasilee - opened

hi,
why not use ChatML as the prompt template?
for a long context it is very easy to have content that contains words/tokens similar to the template you are using now ("User:", "Assistant:")

Because this is not a fine-tuned model. It allows you to run simulated Q/A, Instruct, etc templates, but it's not fine-tuned on them. You'd need a finetune to allow ChatML.

Sign up or log in to comment