zerofata/MS3.2-PaintedFantasy-v2-24B
LoRA-style vs ConicCat/Mistral-Small-3.2-AntiRep-24B. 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: Embeddings 0/1, Norms 0/81, MLP 120/120, Attention 160/160, lm_head 0/1
5 derivatives checked
- bartowski/zerofata_MS3.2-PaintedFantasy-v2-24B-GGUF: Same weights, converted
- mradermacher/MS3.2-PaintedFantasy-v2-24B-i1-GGUF: Quantization
- mradermacher/MS3.2-PaintedFantasy-v2-24B-GGUF: Quantization
- DevQuasar/zerofata.MS3.2-PaintedFantasy-v2-24B-GGUF: Same weights, converted
- zerofata/MS3.2-PaintedFantasy-Visage-v2-33B: Full fine-tune (fewer layers)