mradermacher/Gemma3-4b-KE-strict-r32-full-GGUF
Same weights, converted (fewer layers) vs Samarth2511/Gemma3-4b-KE-strict-r32-full. mradermacher/Gemma3-4b-KE-strict-r32-full-GGUF matches part of Samarth2511/Gemma3-4b-KE-strict-r32-full (444 of its 883 tensors; the rest of Samarth2511/Gemma3-4b-KE-strict-r32-full has no counterpart). On the shared part: Same weights stored in another dtype (BF16 → F16/F32).
Tensors changed: Embeddings 0/1, Norms 0/205, MLP 0/102, Attention 0/136