transformer_blocks: FluxTransformerBlock(174.17 GMACs = 1.75% MACs, 348.4 GFLOPS = 0.88% FLOPs

      (norm1): AdaLayerNormZero(113.25 MMACs = 0% MACs, 232.79 MFLOPS = 0% FLOPs
      (norm1_context): AdaLayerNormZero(113.25 MMACs = 0% MACs, 239.08 MFLOPS = 0% FLOPs
      (attn): Attention(57.98 GMACs = 0.58% MACs, 115.96 GFLOPS = 0.29% FLOPs
      (norm2): LayerNorm(0 MACs = 0% MACs, 6.29 MFLOPS = 0% FLOPs
      (ff): FeedForward(38.65 GMACs = 0.39% MACs, 77.32 GFLOPS = 0.19% FLOPs
      (norm2_context): LayerNorm(0 MACs = 0% MACs, 12.58 MFLOPS = 0% FLOPs
      (ff_context): FeedForward(77.31 GMACs = 0.78% MACs, 154.63 GFLOPS = 0.39% FLOPs



single_transformer_blocks: FluxSingleTransformerBlock(174 GMACs = 1.75% MACs, 348.04 GFLOPS = 0.88% FLOPs
      (norm): AdaLayerNormZeroSingle(56.62 MMACs = 0% MACs, 132.13 MFLOPS = 0% FLOPs
      (proj_mlp): Linear(57.98 GMACs = 0.58% MACs, 115.96 GFLOPS = 0.29% FLOPs
      (act_mlp): GELU(0 MACs = 0% MACs, 18.87 MFLOPS = 0% FLOPs, approximate='tanh')
      (proj_out): Linear(72.48 GMACs = 0.73% MACs, 144.96 GFLOPS = 0.36% FLOPs
      (attn): Attention(43.49 GMACs = 0.44% MACs, 86.97 GFLOPS = 0.22% FLOPs
