Qiming Bao comments

Results 8 comments of


Qiming Bao

Transformers和AllenNLP版本问题

> 你好， AllenNLP目前要求的tansformers 版本 < 3.0 , 不然会报冲突。 > 但是在实际使用的时候，可以正常运行，没有关系。 > 我这里的版本: > > * AllenNLP: 1.0.0 > * transformers: 3.0.2 多谢回复，我试了transformer 3.0.0和AllenNLP 1.0.0虽然有冲突，但是也可以运行。还有一个问题就是目前运行下来模型没有保存下来，然后最后打印的结果是最好的预测结果，不知道后续会不会考虑把训练好的模型保存下来，然后打印结果可以打印平均准确率。谢谢，再次感谢分享。

请问一下博主知道有ELMO+ESIM的组合开源项目吗？

> 你好，不好意思，这个组合没有用过。简单搜了，好像差不多。 AllenNLP应该是有封装ESIM的，如果想用ELMo+ESIM只需要吧input encoder 改了就行的样子。 > 可以尝试： > https://github.com/allenai/allennlp/blob/580dc8b0e2c6491d4d75b54c3b15b34b462e0c67/allennlp/models/esim.py 多谢！

Does Esim with bert performs better than only using bert in text matching task ?

> I used your Esim with bert model on my text matching data set and the performance is worce than only using BERT to finetune. So I wonder the results...

Waiting too long for the rouge assessment in Colab

Oh, Sorry, I just mentioned that if I add --verbose, the console will print out the result. But does anyone knows how long are you waiting for the end of...

An error when I run the remote connection code

Hi, Theamrzaki! Thank you so much for your feedback. I have tried your method, I can connect with my google drive after that, but it seems like the error still...

LayoutLMv3 | Difficulties masking data to do domain adaptation on the base model

Hi, as @NielsRogge said Transfomers have updated the code for mask image modeling and the code is based on DEIT. You can inherit the [code](https://github.com/huggingface/transformers/tree/main/examples/pytorch/image-pretraining#using-datasets-from--datasets) to implement the Mask Image...

what is the evaluation_strategy and how to use it?

It can be selected as `no`, `steps` and `epoch` for the evaluation.

what is the evaluation_strategy and how to use it?

Does the `--evaluation_strategy "steps"` not working? I set this parameter and then I got the following error. How can I add the eval_dataset in the training script? ``` Traceback (most...