cnmoro/Tucano-160m-Portuguese-Instruct-v2
LoRA-style vs TucanoBR/Tucano-160m. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, k_proj, o_proj, q_proj, up_proj, v_proj changed, in 12 of 12 layers; norms, biases and embeddings are untouched.
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/25, MLP 36/36, Attention 48/48
1 derivatives checked
- mradermacher/Tucano-160m-Portuguese-Instruct-v2-GGUF: Same weights, converted