Zexiry/Zera-35B-A3B-4bit
Partial fine-tune vs mlx-community/Qwen3.6-35B-A3B-4bit. Partial change: 1% of the weights changed (attention), in 16 of 40 layers.
Tensors changed: Embeddings 0/3, Norms 0/131, Attention 228/660, MLP 0/960, lm_head 0/3
Partial fine-tune vs mlx-community/Qwen3.6-35B-A3B-4bit. Partial change: 1% of the weights changed (attention), in 16 of 40 layers.
Tensors changed: Embeddings 0/3, Norms 0/131, Attention 228/660, MLP 0/960, lm_head 0/3