Highest on HF Leaderboard!

#2
by LordTwave - opened

Your 7B beats anything else on the HF TruthfulQA leaderboard. Impressive!

The next best on the Leaderboard is a 24B, more computationally expensive than your smaller model.
https://huggingface.co/daxiongshu/Pluto_24B_DPO_63
Pluto_24B_DPO_63;
(Yours) 79.79 - (Theirs) 79.36 = 0.44%

0.44% might not sound like much, but that's huge when you think about it. Your small LLM can beat a larger one on the same widely revered human benchmark, and there's only 20.21% left to be gained on this leaderboard - going the last mile with a small LLM is hugely impressive and deserves a well-done for contributing to the world. Congratulations!

Sign up or log in to comment