Views
No views yet
1from auto_round import AutoRound
2from auto_round.logger import logger as ar_logger
3
4ar = AutoRound(
5 './MiniMax-M2.1-REAP-40',
6 device='cuda',
7 device_map='auto',
8 nsamples=128,
9 seqlen=2048,
10 batch_size=8,
11 enable_torch_compile=False
12)
13ar.quantize_and_save('./MiniMax-M2.1-REAP-40-W4A16', format='auto_round')