Plwy
Plwy
I train tiny on a 8g 2070ti ,batch=16, when epoch=1, it occurs out of memory. image size is 1024*1024.
if you don't have ground truth alpha matte, you need to generate. The mask is just a coarse seg of the person ,can not represent the ground truth alpha matte...
我也遇到了这个问题，使用int8量化后模型能成功转出，但是推理结果除了box坐标值其他值都为0。但是使用混合精度量化或者量化设置为false就没问题。
我遇到了同样的问题，可以将`funasr/auto/auto_model.py`文件的第575行改一下，将result["text"]改为raw_text，像这样 ```python elif self.punc_model is not None: punc_res = self.inference( raw_text, #result["text"], model=self.punc_model, kwargs=self.punc_kwargs, **cfg ) ``` 改一下就可以成功运行了，这样改最简单，但这不是最好的改法，这里punc_model推理了两次导致了文本标点符号多了。
I encountered the same problem.My local status is that libcudnn8.9.6 has been installed, but I encountered this error when running in the conda environment. and solved it using the following...