frank-default-small: ncpu-moe=32(8GPU, ~51.1 t/s gen)

scripts/frank-default-small.env

@@ -31,8 +31,8 @@ NGRAM_MOD_N_MIN=48
 NGRAM_MOD_N_MAX=64
 NGRAM_MOD_N_MATCH=24
 
-# Expert offload: 35 of 40 layers keep MoE weights on CPU; 5 on GPU.
-N_CPU_MOE=35
+# Expert offload: 32 of 40 layers keep MoE weights on CPU; 8 on GPU.
+N_CPU_MOE=32
 N_GPU_LAYERS=all
 
 # KV cache on GPU, both K and V at f16.