avlp12/Laguna-S-2.1-Alis-MLX-Highest-Quality-4.89bpw
Text Generation • 118B • Updated • 251 • 2
Receipt-sealed ALIS-DWQ MLX quants of Laguna S 2.1 (117.6B MoE) for Apple Silicon — highest-quality 4.89 bpw and capacity-optimal 3.31 bpw builds.
Note Quality / highest-fidelity tier: 4.89 bpw, 71.91 GB — 25 accepted DWQ rounds, KL 0.107977. Clean-reproduced on M3 Ultra; actual context tested to 32K.
Note Capacity / size-optimal tier: 3.31 bpw, 48.60 GB — experts 3-bit, attention/dense/shared 4-bit, embeddings/head 6-bit; 24 accepted DWQ rounds. Passed minimum-quality gates; actual context tested to 32K.