{ "current_gpu_type": "NVIDIA A100-PCIE-40GB", "current_gpu_total_memory": 40339.3125, "perplexity": 10.61670970916748, "token_generation_latency_sync": 83.60249481201171, "token_generation_latency_async": 84.0058084577322, "token_generation_throughput_sync": 0.011961365533990301, "token_generation_throughput_async": 0.011903938767557404, "token_generation_CO2_emissions": null, "token_generation_energy_consumption": null, "inference_latency_sync": 108.37442474365234, "inference_latency_async": 108.58149528503418, "inference_throughput_sync": 0.009227269278387303, "inference_throughput_async": 0.009209672397446071, "inference_CO2_emissions": null, "inference_energy_consumption": null }