qyr0403

icon company Renmin University of China icon location Beijing icon location Master’s student at RUC

Results 3 comments of qyr0403

第一个step后第三阶段loss变为nan

> 此外如果我把fp16设置为True会得到 > > ``` > Traceback (most recent call last): > File "tasks/train_it.py", line 213, in > main(cfg) > File "tasks/train_it.py", line 161, in main > global_step = train(...

如何使mvbench推理结果限定为abcd四个选项中的一个？

把其他模型prompt上末尾的特殊token去掉，比如\之类的，就可以了。

Failed to build dropout-layer-norm

have you solve this problem？i also meet this