Wang Ran (汪然)

[Suggestion] Allow to specify multiple GPUs when using `allennlp command` with `--cuda-device`.

3

**Is your feature request related to a problem? Please describe.** Because it is valid to specify multiple GPUs in the config file, I think it is also valid to specify...

use `=` to test equalities of two strings instead of `==`

2

`[ ${TMSPAN} == tag_mspan ]` -> `[ ${TMSPAN} = tag_mspan ]` There may be a bug. (see https://man.linuxde.net/test)

`question_span` is an invalid choice for `answering_ability`.

`initialize B the same way as the default for nn.Linear and A to zero` => `initialize A the same way as the default for nn.Linear and B to zero`

**initialize A the same way as the default for nn.Linear and B to zero** But why not use `nn.init.kaiming_uniform_(self.lora_A, a=math.sqrt(5))` as `class Linear`?

config.yaml是SGM的配置文件还是SGM+GE的配置文件？

1

您好，十分感谢您可以将这份工作的数据及代码开源出来。 :+1: 但是，在使用您的代码过程中，发现了一些小问题 #20 （python train.py时抛出错误）。 另外，我注意到在SGM这篇论文中提到了GE这个组件，且该组件发挥了很重要的作用。但该仓库只提供了一个config.yaml，我不太清楚这个配置是SGM模型的，还是SGM+GE的。我有漏掉什么东西么？您可以同时提供这两个模型的配置文件么？ 再次谢谢您百忙之中回答我的问题。:smile:

just fix error in `numpy_ml.linear_models.rst` from ``` where (\mathbf{X}^\top \mathbf{X} + \alpha \mathbf{I})^{-1} \mathbf{X}^\top is the pseudoinverse / Moore-Penrose inverse adjusted for the L2 penalty on the model coefficients. ```...

Description: --- In the current version, the `COMMAND` definitions related to the CLI have been moved to `cli_config`. This change caused the existing unit tests for `cli_parser` to become outdated...