netcat420/MFANNv0.20.12
LoRA-style vs mlabonne/Meta-Llama-3.1-8B-Instruct-abliterated. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of gate_proj, k_proj, o_proj, q_proj, v_proj changed, in 32 of 32 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/65, MLP 29/96, Attention 128/128, lm_head 0/1
3 derivatives checked
- featherless-ai-quants/netcat420-MFANNv0.20.12-GGUF: Quantization
- mradermacher/MFANNv0.20.12-GGUF: Same weights, converted
- netcat420/MFANNv0.21: LoRA-style