苏剑林(Jianlin Su)

Or any released code to explain how to construct whole word masking for Chinese?

微信多开，每个用户能否独立端口？

2

首先我已经找到了这个答案： https://github.com/TKkk-iOSer/wechat-alfred-workflow/issues/2 然后我想请问一下，能不能给每个用户分配一个独立的端口呢？每个用户独立管理alfred？

只支持huggingface训练的tensorflow模型载入么？

4

首页写到“我们提供了载入huggingface/transformers的pytorch和tensorflow预训练模型方式”，但是 https://github.com/Tencent/TurboTransformers/blob/master/example/python/README_cn.md 又写到“首先我们需要准备一个使用huggingface训练好的bert模型”，也没看到tensorflow方式的例子。

something wrong in Chinese ？

30

in python 2.7: `from flashtext import KeywordProcessor` `keyword_processor = KeywordProcessor()` `keyword_processor.add_keyword(u'北京')` `keyword_processor.add_keyword(u'欢迎')` `keyword_processor.add_keyword(u'你')` `keyword_processor.extract_keywords(u'北京欢迎你')` return [u'北京', u'你']，missing u'欢迎' ?

make a real python2 supporting

1

f-string syntax only supports by python 3.6+. I just replace them with the common format.

I find prefix method in all of your trie modules does not return the postions of the result. For example, datrie module: > > > trie.prefix_items(u'foobarbaz') > > > [(u'foo',...

why not use KL divergence to estimate mutual information

6

As we know I(X,Z) = KL(p(x,z)||p(x)p(z)). So why do you estimate mutual information by JSD rather than KL maximization? f-GAN also gives us KL(p(x)||q(x)) = max_T E_{x~p(x)} [T(x)] - E_{x\sim...

how if we use hinge loss as EnergyModel loss?

3

Good job! But I have some problems: 1. how if we use hinge loss as EnergyModel loss? Gradient penalty is actually good but we know with gradient penalty is slow....

how can we explain GAN works without I(X,Z) term?

2

compared with wgan-gp or wgan-div, your new GAN has an additional I(X,Z) term on generator loss. This term may prevent generator from mode collapse. As we known, wgan-gp or wgan-div...

Does not support this kind of Gather

2

**Please describe the bug** I attempted to use alpa, and the quick start example works fine. However, I encountered an error when implementing it with my own model. My model...