shisa-ai/ablation-87-bestofn.atlo.rp.tl.unphi42e6-shisa-v2-unphi-4-14b
Full fine-tune vs unsloth/phi-4. Fine-tune: 100% of the weights changed, across 40 of 40 layers.
Tensors changed: Embeddings 1/1, Norms 12/81, MLP 120/120, Attention 160/160, lm_head 1/1