auto-round
add gptqmodel 6.0 compatible change and MTP quantization skip warning.
#1663
Merged
Merged