UnfilteredAI/BADMISTRAL-1.5B
LoRA-style vs OEvortex/BabyMistral. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, k_proj, o_proj, q_proj, up_proj, v_proj changed, in 20 of 20 layers; norms, biases and embeddings are untouched.
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/41, MLP 60/60, Attention 80/80
4 derivatives checked
- MaziyarPanahi/BADMISTRAL-1.5B-GGUF: Same weights, converted
- mradermacher/BADMISTRAL-1.5B-GGUF: Quantization
- Novaciano/BADMISTRAL-1.5B-NSFW2-GGUF: Quantization
- mradermacher/BADMISTRAL-1.5B-i1-GGUF: Quantization