unsloth/gemma-4-E4B-it-qat-q4_0-unquantized
Full fine-tune vs google/gemma-4-E4B-it. Fine-tune: 94% of the weights changed, across 42 of 42 layers.
Tensors changed: MLP 126/606, Norms 212/484, other 126/261, Attention 132/719, Embeddings 3/6
4 derivatives checked
- mlx-community/unsloth-gemma-4-E4B-it-qat-oQ4: Quantization
- efficiencyx/Jun-LoRA-E4B-GGUF: LoRA-style (fewer layers)
- efficiencyx/Jun-v4-E4B-Step100-GGUF: LoRA-style (fewer layers)
- tbhrc/unsloth_gemma_4_e4b_it_qat_oq4: Quantization
4 other repos hold byte-for-byte identical weight files.