ConicCat/Role-mo-V2-32B
LoRA-style vs allenai/Olmo-3.1-32B-Instruct. 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 64 of 64 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, MLP 192/192, Norms 0/257, Attention 256/256, lm_head 0/1
2 derivatives checked
- mradermacher/Role-mo-V2-32B-i1-GGUF: Quantization
- mradermacher/Role-mo-V2-32B-GGUF: Quantization