Jason Phang comments

Results 35 comments of


                                            Jason Phang

PEGASUS-X

I can follow up on the rest of the feedback this weekend / early next week: most of it looks manageable. One comment on `DimensionInfo`: I use it to capture...

PEGASUS-X

Let me know if there is anything else I need to address!

PEGASUS-X

Let me ping Peter Liu on this. He should be able to pull and push to the Google org. I will update the paths in the PR when it is...

Unable to run RAG model for tokenizing, fine-tuning, and predicting with `jiant` on branch `IRT_experiments`

Hi, sorry for the delay in my response. * `normalize_tokenizations` has to do with aligning token spans between raw text and the model tokenizer's tokens. Depending on which tokenizer you're...

How can I reproduce the results in SuperGlue paper?

The results for SuperGLUE were computed using a legacy version of the `jiant` repository, found here: https://github.com/nyu-mll/jiant-v1-legacy

Unable to reproduce XTREME numbers

Hi, I believe the issue may have been that the XLM-R weights not being correctly loaded because of a recent update. I've made a PR that should address the issue...

Will you support T5 or GPT based models in the future

We have yet to considered T5 because of the memory requirements of the largest models, but it should be feasible to incorporate within the current framework. We have also considered...

Will you support T5 or GPT based models in the future

Hi, This repo is no longer being actively maintained, so we have no plans to add support for T5 and GPT-2 now.

Error during edge-probing task when a span is located beyond the max_seq_length

Thanks! I will take a look at this.

Custom torch.utils.data.DataLoader.collate_fn

Hi, Not as far as I know. Feel free to make a PR!