VVNMA
Results
2
comments of
VVNMA
Hi, Did you solve this problem? I met the same problem as yours. I use a machine with 8*p100 16G, it did not cause OOM.
@DachengLi1 Thanks. I saw the replace in the model adapter. But for the gradio server, I have to modify the input and max output token to fit the long chat.