EPFLiGHT/Apertus-70B-MeditronFO

Full fine-tune vs swiss-ai/Apertus-70B-Instruct-2509. Fine-tune: 100% of the weights changed, across 80 of 80 layers.

Tensors changed: Embeddings 1/1, Norms 92/321, MLP 170/480, Attention 320/320, lm_head 1/1

2 derivatives checked