jaspionjader/Kosmos-EVAA-Franken-v36-8B

Partial fine-tune vs jaspionjader/f-5-8b. Partial change: 65% of the weights changed (attention, mlp, norms), in 24 of 32 layers.

Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 48/65, MLP 72/96, Attention 96/128

3 derivatives checked