blascotobasco/LFM2.5-4E-A2.4B-Abliterated
LoRA-style vs huihui-ai/Huihui-LFM2.5-1.2B-Instruct-abliterated. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of k_proj, q_proj, v_proj, w1, w2, w3 changed, in 8 of 16 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/2, other 0/30, MLP 6/6, Norms 0/44, Attention 18/24
2 derivatives checked
- mradermacher/LFM2.5-4E-A2.4B-Abliterated-i1-GGUF: Quantization
- mradermacher/LFM2.5-4E-A2.4B-Abliterated-GGUF: Same weights, converted