Vimos Tan

``` diff --git a/requirements.txt b/requirements.txt index ff637fc..fd111d7 100755 --- a/requirements.txt +++ b/requirements.txt @@ -4,5 +4,5 @@ Flask theano keras lasagne -ntlk +nltk //nltk.download('punkt') ```

出错如下 ``` BibTeX: Too many commas in name 1 of "Liang Xu, Xuanwei Zhang, Lu Li, Hai Hu, Chenjie Cao, Weitang Liu, Junyi Li, Yudong Li, Kai Sun, Yechen Xu,...

The project is well-documented in Chinese, is there any plan to support English documentation?

Use Roberta in pytorch transformers

5

I have been using `BertPreTrainedModel` to load this roberta model, which works well. Noticing in `pytorch_transformers`, `Roberta` is also supported. ``` from pytorch_transformers import (BertConfig, BertTokenizer, RobertaConfig, RobertaTokenizer) ``` Should...

「变调」“不”字的标音问题

5

基本信息 只有一个“不”字或“不”字后面紧接着的字为非第四声时，读音为四声bù； 例如：不（bù）法（fǎ）之徒 但是如果后面的字也是四声，需要变调，变成bú。 例如：不（bú）是；不（bú）对 但从标音角度讲，变调不应作为一种标音规则，不（bù）才是正确的标音。

Confused by multiply n to both clf_logits and clf_losses?

3

I am confused by the code below. https://github.com/huggingface/pytorch-openai-transformer-lm/blob/eafc28abdfadfa0732f03a0fc65805c5bfb2ffe7/train.py#L52 https://github.com/huggingface/pytorch-openai-transformer-lm/blob/eafc28abdfadfa0732f03a0fc65805c5bfb2ffe7/train.py#L54 Is this due to any normalization? Thanks!

The pdf links of RANLP2021 papers are not a concatenation of the page link and the extension `.pdf`. For example, the page link [https://aclanthology.org/2021.ranlp-1.156/](https://aclanthology.org/2021.ranlp-1.156/), adding `.pdf` [https://aclanthology.org/2021.ranlp-1.156.pdf](https://aclanthology.org/2021.ranlp-1.156.pdf) does not exist...

I tried to run the code, it seems needs a long time. I think it would be helpful to post some experiment results. So that we can compare with the...

Just curious, is it common practice to add extra embedding to BERT but keep using original encoder parameters?

I noticed that you used many transposes rather than one permutation in the flow operation, is that for better performance? I think the following code is equivalent to the original...