SvalTek/Llama3.2-ColdBrew-4B-Base
LoRA-style vs bunnycore/Llama-3.2-3B-RP-DeepThink. 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 28 of 28 layers; norms, biases and embeddings are untouched.
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/57, MLP 84/84, Attention 112/112
2 derivatives checked
- mradermacher/Llama3.2-ColdBrew-4B-Base-GGUF: Same weights, converted
- SvalTek/Llama3.2-ColdBrew-4B-Thinker: Full fine-tune