Comment by tagawa

Repo with benchmark tasks, evaluation harness, tech report:

https://github.com/perplexityai/wandr