{ "current_gpu_type": "NVIDIA A100-PCIE-40GB", "current_gpu_total_memory": 40339.3125, "perplexity": 7.51709508895874, "token_generation_latency_sync": 54.72829818725586, "token_generation_latency_async": 54.37298659235239, "token_generation_throughput_sync": 0.018272082873442282, "token_generation_throughput_async": 0.018391485601062255, "token_generation_CO2_emissions": null, "token_generation_energy_consumption": null, "inference_latency_sync": 193.2348403930664, "inference_latency_async": 108.21371078491211, "inference_throughput_sync": 0.005175050202985454, "inference_throughput_async": 0.009240973188578862, "inference_CO2_emissions": null, "inference_energy_consumption": null }