Lei Xu

* Orion version: Commit 23f2bcc * Python version: 3.6 * Operating System: Google Colab ### Description I tried to use the colab notebook. I think it's nice and clear. Here...

Maybe a bug in constructing hidden states.

2

``` (1) hidden = self.latent2hidden(z) (2) if self.bidirectional or self.num_layers > 1: (3) hidden = hidden.view(self.hidden_factor, batch_size, self.hidden_size) (4) else: (5) hidden = hidden.unsqueeze(0) ``` I think line (3) should...

Higher LF accuracy on sample output.

2

Hi, I'm new to this data set. I followed README and run evaluation on example.pred.dev.jsonl. I got the following result. { "ex_accuracy": 0.5380596128725804, "lf_accuracy": 0.45208407552547203 } I'm not sure what...

- Fix all docstrings. - Evaluator and FeatureToolsWrapper - Give a clear explanation about - Fix Load and Store for prediction problems - Used to store executable code for cutoff...

The functionality of DatasetForBert is becoming more complicated. It's good to have a unit test for that module.

Current Dataset manually constructs the input tensor, masks, etc. We can make it more clear by using transformers.BertTokenizerFast

Add a wrapper to support all datasets from huggingface/datasets library.

About OU Noise Function

1

I think in the paper, they use a modified OU process in section 8.2. I'm not sure if that is implemented in your code?