suayptalha/Qwenhancement-32B-v1

Partial fine-tune vs Sakalti/ultiima-32B. Partial change: 59% of the weights changed (attention, lm_head, mlp, norms), in 64 of 64 layers.

Tensors changed: lm_head 1/1, Embeddings 0/1, Norms 46/129, MLP 110/192, Attention 267/448

2 derivatives checked