ResearchBenchmarksEfficiency Evaluation for Generative Recommendation on SteamEfficiency Evaluation for Generative Recommendation on SteamFollow0.07FLOPs (G)Fave-0.0080.51851.0451.5715Apr 6, 2026Evaluation resultsMethodMethodLinksFLOPs (G)Latency (ms)Inference Time (s)FaveHardware=NVIDIA L40 GPUHardware=NVIDIA L40 GPU2026.040.073.9926.23DiffuRecHardware=NVIDIA L40 GPUHardware=NVIDIA L40 GPU2026.041.26104.45507.23FMRecHardware=NVIDIA L40 GPUHardware=NVIDIA L40 GPU2026.042.0292.81516.27