v1.2 merged back to LunarMix to fix overfitting
This is the finetuned version of LunarMix, trained with OFTv2. An OFTv2 model was trained on LunarMix. The resulting model is then merged to LunarMix.
