mlx-community/gemma-3n-E4B-it-lm-bf16
Same weights, repacked (fewer layers) vs google/gemma-3n-E4B-it. mlx-community/gemma-3n-E4B-it-lm-bf16 matches part of google/gemma-3n-E4B-it (851 of its 1676 tensors; the rest of google/gemma-3n-E4B-it has no counterpart). On the shared part: Same weights in different files (re-saved, re-sharded or new metadata).
Tensors changed: MLP 0/108, Embeddings 0/5, other 0/281, Norms 0/317, Attention 0/140