SicariusSicariiStuff/Impish_Magic_24B
LoRA-style vs mistralai/Magistral-Small-2506. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, k_proj, o_proj, q_proj, up_proj changed, in 40 of 40 layers; norms, biases and embeddings are untouched.
Tensors changed: Norms 0/81, MLP 42/120, Attention 47/160
3 derivatives checked
- mradermacher/Impish_Magic_24B-i1-GGUF: Quantization
- SicariusSicariiStuff/Impish_Magic_24B_GGUF: Quantization
- mradermacher/Impish_Magic_24B-GGUF: Quantization