YOYO-AI/YOYO-O1-14B-V2

Full fine-tune vs FractalAIResearch/Fathom-R1-14B. Fine-tune: 100% of the weights changed, across 48 of 48 layers.

Tensors changed: Norms 97/97, MLP 144/144, Attention 336/336

2 derivatives checked