Hongbin Yang

@daiab （不知是否扯远了。）之前我还不知有MXNET这个框架，感谢提出，观察了下里面教程非常好，比如http://mxnet.io/architecture/program_model.html 描述了Symbolic Programs，我才想通了之前以前无法理解的Tensor的概念~ 个人感觉从用法上此框架更像TensorFlow，不过我猜测真正的计算引擎和TensorFlow、MXNET都不一样吧，毕竟他们都是C++写的。

Hi, I also want to train the model in parallel. How to implement `nn.DataParallel()`? Once I set that `transformer=nn.Dataparallel(transformer).to(device)` in `train.py`, the training process then kept in 0% and could...

Hi @jnwei I wonder is it possible to first feed a large dataset such as ZINC without the properties and then feed a much small one with the property such...

@seaspeak https://github.com/sparanoid/chinese-copywriting-guidelines

I think the interpreter may be wrong. I mean this is not in a python runtime. Please try: ` python data_structs.py`. First you should make sure your python environment is...

数字后面跟英文字符的前提是他们是两个单词，如数字加单位组合。但你无法保证他们在书写时是两个单词的意思。除了st/rd/th这种可以预设的外，还有比如 “i18n” 这种数字+英文组合亦或者是用户想要输入些代码，如服务器型号之类的。 我建议还是不要自动化。