UncannyEcho/Aura-Large-v1-BF16
LoRA-style vs google/gemma-4-31B-it. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, k_proj, o_proj, q_proj, v_proj changed, in 60 of 60 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/4, Norms 0/523, other 0/62, MLP 48/261, Attention 213/338
2 derivatives checked
- mradermacher/Aura-Large-v1-BF16-i1-GGUF: Quantization (fewer layers)
- mradermacher/Aura-Large-v1-BF16-GGUF: Quantization (fewer layers)