mratsim/MiniMax-M2.5-BF16-INT4-AWQ · Discussions

Resources

View closed (0)

vLLM's log reports a 'incorrect tokenization'.

#5 opened 1 day ago by

accuracy

#4 opened 2 days ago by

accuracy benchmark

#3 opened 2 days ago by

FP8 + INT4 version

#2 opened 2 days ago by

Cant get it to work on 8x RTX3090

#1 opened 2 days ago by