franciscobdl/Granite4-Estigia-MultilanguageV1
LoRA-style vs unsloth/granite-4.0-h-tiny. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of input_linear, k_proj, o_proj, output_linear, q_proj, v_proj changed, in 40 of 40 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, other 0/372, Norms 0/117, MLP 80/80, Attention 16/16
1 derivatives checked
- mradermacher/Granite4-Estigia-MultilanguageV1-GGUF: Same weights, converted