huggingface-KREW/EXAGIRL-2.4B-Instruct
LoRA-style vs LGAI-EXAONE/EXAONE-3.5-2.4B-Instruct. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of k_proj, q_proj, v_proj changed, in 30 of 30 layers; norms, biases and embeddings are untouched.
Tensors changed: Attention 90/120, Norms 0/61, MLP 0/90, Embeddings 0/1
1 derivatives checked
- mradermacher/EXAGIRL-2.4B-Instruct-GGUF: Same weights, converted