{ "current_gpu_type": "NVIDIA A100-PCIE-40GB", "current_gpu_total_memory": 40339.3125, "perplexity": 10.041903495788574, "token_generation_latency_sync": 60.611341094970705, "token_generation_latency_async": 60.60555353760719, "token_generation_throughput_sync": 0.016498562512139766, "token_generation_throughput_async": 0.01650013805054146, "token_generation_CO2_emissions": null, "token_generation_energy_consumption": null, "inference_latency_sync": 191.5773941040039, "inference_latency_async": 108.57903957366943, "inference_throughput_sync": 0.005219822540529589, "inference_throughput_async": 0.009209880690844694, "inference_CO2_emissions": null, "inference_energy_consumption": null }