lmstudio-community/gemma-4-E4B-it-QAT-GGUF
Quantization (fewer layers) vs google/gemma-4-E4B-it-qat-q4_0-unquantized. lmstudio-community/gemma-4-E4B-it-QAT-GGUF matches part of google/gemma-4-E4B-it-qat-q4_0-unquantized (623 of its 2076 tensors; the rest of google/gemma-4-E4B-it-qat-q4_0-unquantized has no counterpart). On the shared part: lmstudio-community/gemma-4-E4B-it-QAT-GGUF is a GGUF quantization (Q4_0) of the same weights: 279 tensors match exactly and 344 within quantization error.
Tensors changed: Embeddings 0/2, Norms 0/278, MLP 0/126, other 0/85, Attention 0/132