aiuser3993/DefCPM-2B
LoRA-style vs openbmb/MiniCPM5-2B-Base. 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 42 of 42 layers; norms, biases and embeddings are untouched.
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/85, MLP 126/126, Attention 168/168
2 derivatives checked
- mradermacher/DefCPM-2B-i1-GGUF: Quantization
- mradermacher/DefCPM-2B-GGUF: Same weights, converted