SubMaroon/Boulesis-v2.1-26B-A4B-HeadTune

Full fine-tune vs Gryphe/Pantheon-Reasoning-26B-A4B-1.1-V2. Fine-tune: 94% of the weights changed, across 30 of 30 layers.

Tensors changed: lm_head 1/1, Embeddings 0/4, MLP 149/231, Norms 0/433, other 30/122, Attention 115/223

1 derivatives checked