madoss/LFM2-2.6B-FRMOO-V2
LoRA-style vs LiquidAI/LFM2-2.6B. 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 8 of 30 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/2, other 0/66, MLP 0/90, Norms 0/76, Attention 24/32
1 derivatives checked
- mradermacher/LFM2-2.6B-FRMOO-V2-GGUF: Same weights, converted