{ "current_gpu_type": "NVIDIA A100-PCIE-40GB", "current_gpu_total_memory": 40339.3125, "perplexity": 8.666143417358398, "token_generation_latency_sync": 163.67237854003906, "token_generation_latency_async": 163.9625133946538, "token_generation_throughput_sync": 0.006109766406036377, "token_generation_throughput_async": 0.0060989550556170375, "token_generation_CO2_emissions": null, "token_generation_energy_consumption": null, "inference_latency_sync": 260.6235656738281, "inference_latency_async": 192.3330783843994, "inference_throughput_sync": 0.003836951571952268, "inference_throughput_async": 0.005199313651089112, "inference_CO2_emissions": null, "inference_energy_consumption": null }