wenzani/Wendi-2B
LoRA-style vs Qwen/Qwen3.5-2B. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of in_proj_qkv, q_proj, v_proj changed, in 24 of 24 layers; norms, biases and embeddings are untouched.
Tensors changed: Attention 30/264, Norms 0/177, Embeddings 0/4, MLP 0/172
2 derivatives checked
- mradermacher/Wendi-2B-GGUF: Same weights, converted
- wenzani/Wendi-2B-GGUF: Quantization