Alisa Liu

Hi Andrea, Thank you for the response! To clarify, does that mean that an example is counted as a positive example (i.e. toxic) if any one of the toxicity subcategories...

Hi Andrea, Thank you for your help!! I have trained a discriminator with a larger embedding size compatible with GPT2-large, and am able to use this discriminator to steer generations...

Hey @Kadam-Tushar, thanks for the fix! This codebase was based on `transformers==3.3.1`, and I believe this error may be due to a [breaking change](https://huggingface.co/docs/transformers/migration#4-switching-the-returndict-argument-to-true-by-default) from `transformers 3.x` to `4.x` with...

Yes, this part of the evaluation is unfortunately no longer reproducable due to OpenAI's deprecation of GPT-3-based models. The [allenai/open-instruct](https://github.com/allenai/open-instruct) evaluation frameowrk has switched to finetuned judge models based on...