{ "current_gpu_type": "NVIDIA A100-PCIE-40GB", "current_gpu_total_memory": 40339.3125, "perplexity": 150931.8125, "token_generation_latency_sync": 175.739599609375, "token_generation_latency_async": 175.4506889730692, "token_generation_throughput_sync": 0.005690237158971279, "token_generation_throughput_async": 0.0056996071423435395, "token_generation_CO2_emissions": null, "token_generation_energy_consumption": null, "inference_latency_sync": 273.39837646484375, "inference_latency_async": 235.61007976531982, "inference_throughput_sync": 0.0036576661973286803, "inference_throughput_async": 0.00424430058763213, "inference_CO2_emissions": null, "inference_energy_consumption": null }