lianzhaoy

Results 6 comments of lianzhaoy

好的,非常感谢您的答复!代码里面调用了save,但是并没有生成对应的实体对齐结果,请问是否能更新一下代码? ![image](https://user-images.githubusercontent.com/33119533/111724345-0f6eda00-88a0-11eb-9d6d-cee16661bf46.png) ![image](https://user-images.githubusercontent.com/33119533/111724390-1bf33280-88a0-11eb-9e6a-80b6f58443ec.png)

改成tanh后,name_mat = preprocessing.normalize(name_mat)这一句会报错:ValueError: Input contains NaN, infinity or a value too large for dtype('float64'). 请问是由于什么原因造成的呢?

感谢提供测试思路! 请问有相应的测试脚本吗?

求教如何推理单个样本? 自己按照训练代码实现了单个样本推理脚本,对于相同数据,加载两次模型,推理结果不一致是什么问题?

@weberrr Use export_hf_checkpoint.py to merge the basic model and lora weights, and then fine-tune the merged model based on uniform_finetune.py. Is that what it means?