google/gemma-4-E4B-it-qat-q4_0-unquantized-assistant
Full fine-tune vs google/gemma-4-E4B-it-assistant. Fine-tune: 99% of the weights changed, across 4 of 4 layers.
Tensors changed: Embeddings 2/3, Norms 17/21, other 6/6, MLP 12/12, Attention 8/8
8 derivatives checked
- cascade-tech/gemma-4-E4B-it-qat-q4_0-unquantized-assistant-gguf: Same weights, converted
- mradermacher/gemma-4-E4B-it-qat-q4_0-unquantized-assistant-GGUF: Quantization
- amaranus/Gemma-4-E4B-it-qat-assistant-MTP-Q8_0-GGUF: Quantization
- mlx-community/gemma-4-E4B-it-qat-assistant-4bit: Quantization
- mlx-community/gemma-4-E4B-it-qat-assistant-bf16: Same weights, converted
- ensky0/gemma-4-E4B-it-qat-q4_0-gguf-assistant: Quantization
- NicklausCairns/gemma-4-E4B-it-qat-assistant-MTP-Q8_0: Full fine-tune
- manojbillionaire123/Gemma-4-E4B-it-qat-assistant-MTP-Q8_0-GGUF: Quantization
2 other repos hold byte-for-byte identical weight files.