mradermacher/gemma-4-e2b-arabic-english-vision-GGUF
Same weights, converted (fewer layers) vs ml-intern-explorers/gemma-4-e2b-arabic-english-vision. mradermacher/gemma-4-e2b-arabic-english-vision-GGUF matches part of ml-intern-explorers/gemma-4-e2b-arabic-english-vision (565 of its 1259 tensors; the rest of ml-intern-explorers/gemma-4-e2b-arabic-english-vision has no counterpart). On the shared part: Same weights stored in another dtype (BF16 → F16/BF16), plus 36 new tensors.
Tensors changed: Embeddings 0/2, Norms 0/247, MLP 0/105, other 0/71, Attention 0/140