DataPilot/ArrowNeo-AME-4x3B-v0.1-MoE
MoE built from dense models vs sbintuitions/sarashina2.2-3b-instruct-v0.1. DataPilot/ArrowNeo-AME-4x3B-v0.1-MoE is a mixture of 3 experts with the same shape as sbintuitions/sarashina2.2-3b-instruct-v0.1's MLP, but none of them is a copy of it: they come from other models or were trained further. Shared tensors (attention, norms, embeddings): 195 of 195 identical.
2 derivatives checked
- mradermacher/ArrowNeo-AME-4x3B-v0.1-MoE-i1-GGUF: Quantization
- mradermacher/ArrowNeo-AME-4x3B-v0.1-MoE-GGUF: Same weights, converted
1 other repos hold byte-for-byte identical weight files.