Request for Llama 3.1 8B Instruct Evaluation Metrics (Majority Voting, Best-of-N, Beam Search, DVTS) #56
Description
Hello, and thanks for this fantastic project!
I’m interested in the evaluation results for Llama 3.1 8B Instruct using Majority Voting, Best-of-N, Beam Search, and DVTS. I didn’t see those metrics in the blog or README.
Running Llama 3.1 8B Instruct requires a lot of GPU memory and time, which makes it difficult for individual researchers to reproduce these results. Would it be possible to include those metrics in the blog post or the project’s README?
Thank you very much!