mrgnw/gemma-4-e2b-svelte5
Full fine-tune vs mlx-community/gemma-4-e2b-it-4bit. Fine-tune: 95% of the weights changed, across 35 of 35 layers.
Tensors changed: Embeddings 3/6, other 172/245, MLP 219/315, Norms 0/132, Attention 211/300
Full fine-tune vs mlx-community/gemma-4-e2b-it-4bit. Fine-tune: 95% of the weights changed, across 35 of 35 layers.
Tensors changed: Embeddings 3/6, other 172/245, MLP 219/315, Norms 0/132, Attention 211/300