{ "current_gpu_type": "NVIDIA A100-PCIE-40GB", "current_gpu_total_memory": 40339.3125, "perplexity": 8.201188087463379, "token_generation_latency_sync": 18.678534698486327, "token_generation_latency_async": 18.708267621695995, "token_generation_throughput_sync": 0.053537390172315714, "token_generation_throughput_async": 0.05345230356018101, "token_generation_CO2_emissions": null, "token_generation_energy_consumption": null, "inference_latency_sync": 184.60794830322266, "inference_latency_async": 49.54235553741455, "inference_throughput_sync": 0.00541688485891993, "inference_throughput_async": 0.020184748770065982, "inference_CO2_emissions": null, "inference_energy_consumption": null }