Kasper-Bankler/gemma-4-E2B-Alignment-Study

Full fine-tune vs google/gemma-4-E2B. Fine-tune: 91% of the weights changed, across 35 of 35 layers.

Tensors changed: MLP 105/585, Norms 177/433, other 105/240, Attention 100/687, Embeddings 4/6

2 derivatives checked

• mradermacher/gemma-4-E2B-Alignment-Study-i1-GGUF: Quantization (fewer layers)
• mradermacher/gemma-4-E2B-Alignment-Study-GGUF: Same weights, converted (fewer layers)

On the Hub · Downloads on Model Pulse