SkyAsl/Nanbeige4.1-VLM
LoRA-style vs Nanbeige/Nanbeige4.1-3B. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of k_proj, o_proj, q_proj, v_proj changed, in 32 of 32 layers; norms, biases and embeddings are untouched.
Tensors changed: MLP 0/96, Attention 128/128, Norms 0/65
1 derivatives checked
- mradermacher/Nanbeige4.1-VLM-GGUF: Same weights, converted