SC117/Laguna-S-2.1-Uncensored
LoRA-style vs poolside/Laguna-S-2.1. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate, o_proj changed, in 24 of 48 layers; norms, biases and embeddings are untouched.
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/193, MLP 284/2565, Attention 23/240
4 derivatives checked
- SC117/Laguna-S-2.1-Uncensored-APEX-GGUF: Full fine-tune
- mradermacher/Laguna-S-2.1-Uncensored-i1-GGUF: Quantization
- mradermacher/Laguna-S-2.1-Uncensored-GGUF: Quantization
- AmixDigital/Laguna-S-2.1-Uncensored-oQ4e: Quantization