Lambent/Arsenic-Shahrazad-12B-v4.4.2-mtp
LoRA-style vs Lambent/Arsenic-Shahrazad-12B-v4.4.1-mtp. 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 40 of 40 layers; norms, biases and embeddings are untouched.
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/81, MLP 120/120, Attention 160/160
3 derivatives checked
- mradermacher/Arsenic-Shahrazad-12B-v4.4.2-mtp-i1-GGUF: Quantization
- mradermacher/Arsenic-Shahrazad-12B-v4.4.2-mtp-GGUF: Quantization
- Lambent/Arsenic-Shahrazad-12B-v4.4.3-mtp: LoRA-style