{ "current_gpu_type": "NVIDIA A100-PCIE-40GB", "current_gpu_total_memory": 40339.3125, "perplexity": 7.073419570922852, "token_generation_latency_sync": 19.17551956176758, "token_generation_latency_async": 19.6293156594038, "token_generation_throughput_sync": 0.05214982555121031, "token_generation_throughput_async": 0.05094421106427777, "token_generation_CO2_emissions": null, "token_generation_energy_consumption": null, "inference_latency_sync": 160.10240020751954, "inference_latency_async": 42.252469062805176, "inference_throughput_sync": 0.006246002550266782, "inference_throughput_async": 0.023667255954050254, "inference_CO2_emissions": null, "inference_energy_consumption": null }