TheDrummer/Magidonia-24B-v4.2.0
LoRA-style (fewer layers) vs mistralai/Magistral-Small-2509. TheDrummer/Magidonia-24B-v4.2.0 matches part of mistralai/Magistral-Small-2509 (363 of its 585 tensors; the rest of mistralai/Magistral-Small-2509 has no counterpart). On the shared part: 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
3 derivatives checked
- bartowski/TheDrummer_Magidonia-24B-v4.2.0-GGUF: Same weights, converted
- mradermacher/Magidonia-24B-v4.2.0-i1-GGUF: Quantization
- mradermacher/Magidonia-24B-v4.2.0-GGUF: Quantization