Update README.md

README.md

@@ -8,7 +8,7 @@ tags:
 - merge
 license: llama3
 ---
-### Yoloing this 99-layer llamastein straight to the leaderboard wish me luck
+### The following is a very random experiment in mixing together the brains of 2 very different finetunes, llama3-70b-instruct, and nous-hermes-pro-70b which was not based on the instruct, but on the base mode, so these 2 models, are actually quite diverged from their common ancesotr, and honestly i wasnt expecting to work well at all, just yoloed it, there's still issues with hallucination due to prompt templates, but overall, end results were very surprising, it's both smart and tarded at once. Most surprising was the 1 bit quant qhich ended up being very coherent.
 
 > [!TIP]
 >for 1bit quant download: