Tushar Khot

Added new HF datasets => Perplexity format functions

1

New datasets: arc-easy, boolqa, obqa, hellaswag, mmlu, xsum, imdb. Uploaded converted files to: [s3://ai2-llm/eval-data/downstream/v0/](https://s3.console.aws.amazon.com/s3/buckets/ai2-llm?region=us-east-1&prefix=eval-data/downstream/v0/) @IanMagnusson for visibility. Feel free to merge whenever convenient.

Download the data to the docker image

7

@HarshTrivedi The image is missing the required data. ``` 2024-10-09T21:20:08.493624068Z [1;91mException: [0mDid not find any .[35m/[0m[95mdata[0m in the AppWorld's root directory which is set 2024-10-09T21:20:08.493628818Z to [32m'.'[0m. Make sure you...

Updated code to call o1

3

@PeterAJansen Updated the README and code to call o1 models. The code works but it will need a lot more tokens to produce an output.

Fix issue with typing_extension and faiss

1

Hi I was facing the same issue as https://github.com/WooooDyy/AgentGym/issues/22 Rather than re-installing numpy, I found the following fix to work. Specifically, I had to fix the faiss-cpu version to `1.7.0`...