Crystalcareai/CrystalMistral-24B
MoE built from dense models vs QuixiAI/WestLake-7B-v2-laser. Crystalcareai/CrystalMistral-24B is a mixture of 4 experts assembled from dense models (mergekit-moe style): expert 3 is an exact copy of QuixiAI/WestLake-7B-v2-laser's MLP; the other experts come from other models. Shared tensors (attention, norms, embeddings): 0 of 195 identical.
2 derivatives checked
- mradermacher/CrystalMistral-24B-i1-GGUF: Quantization
- mradermacher/CrystalMistral-24B-GGUF: Quantization