Danielbrdz/Barcenas-h-1b
LoRA-style vs ibm-granite/granite-4.0-h-1b. 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, Norms 0/117, other 0/252, MLP 80/80, Attention 16/16
2 derivatives checked
- mradermacher/Barcenas-h-1b-i1-GGUF: Quantization
- mradermacher/Barcenas-h-1b-GGUF: Same weights, converted