IlyasMoutawwakil's picture
Upload cuda_inference_transformers_token-classification_microsoft/deberta-v3-base/benchmark.json with huggingface_hub
f94f995 verified
raw
history blame
8.79 kB
{
"config": {
"name": "cuda_inference_transformers_token-classification_microsoft/deberta-v3-base",
"backend": {
"name": "pytorch",
"version": "2.5.1+cu124",
"_target_": "optimum_benchmark.backends.pytorch.backend.PyTorchBackend",
"task": "token-classification",
"library": "transformers",
"model_type": "deberta-v2",
"model": "microsoft/deberta-v3-base",
"processor": "microsoft/deberta-v3-base",
"device": "cuda",
"device_ids": "0",
"seed": 42,
"inter_op_num_threads": null,
"intra_op_num_threads": null,
"model_kwargs": {},
"processor_kwargs": {},
"no_weights": true,
"device_map": null,
"torch_dtype": null,
"eval_mode": true,
"to_bettertransformer": false,
"low_cpu_mem_usage": null,
"attn_implementation": null,
"cache_implementation": null,
"autocast_enabled": false,
"autocast_dtype": null,
"torch_compile": false,
"torch_compile_target": "forward",
"torch_compile_config": {},
"quantization_scheme": null,
"quantization_config": {},
"deepspeed_inference": false,
"deepspeed_inference_config": {},
"peft_type": null,
"peft_config": {}
},
"scenario": {
"name": "inference",
"_target_": "optimum_benchmark.scenarios.inference.scenario.InferenceScenario",
"iterations": 1,
"duration": 1,
"warmup_runs": 1,
"input_shapes": {
"batch_size": 2,
"sequence_length": 16,
"num_choices": 2
},
"new_tokens": null,
"memory": true,
"latency": true,
"energy": true,
"forward_kwargs": {},
"generate_kwargs": {
"max_new_tokens": 2,
"min_new_tokens": 2
},
"call_kwargs": {
"num_inference_steps": 2
}
},
"launcher": {
"name": "process",
"_target_": "optimum_benchmark.launchers.process.launcher.ProcessLauncher",
"device_isolation": true,
"device_isolation_action": "error",
"numactl": false,
"numactl_kwargs": {},
"start_method": "spawn"
},
"environment": {
"cpu": " AMD EPYC 7R32",
"cpu_count": 16,
"cpu_ram_mb": 66697.248768,
"system": "Linux",
"machine": "x86_64",
"platform": "Linux-5.10.228-219.884.amzn2.x86_64-x86_64-with-glibc2.35",
"processor": "x86_64",
"python_version": "3.10.12",
"gpu": [
"NVIDIA A10G"
],
"gpu_count": 1,
"gpu_vram_mb": 24146608128,
"optimum_benchmark_version": "0.5.0.dev0",
"optimum_benchmark_commit": null,
"transformers_version": "4.47.0",
"transformers_commit": null,
"accelerate_version": "1.2.0",
"accelerate_commit": null,
"diffusers_version": "0.31.0",
"diffusers_commit": null,
"optimum_version": null,
"optimum_commit": null,
"timm_version": "1.0.12",
"timm_commit": null,
"peft_version": "0.14.0",
"peft_commit": null
},
"print_report": true,
"log_report": true
},
"report": {
"load_model": {
"memory": {
"unit": "MB",
"max_ram": 815.857664,
"max_global_vram": 1410.859008,
"max_process_vram": 0.0,
"max_reserved": 773.849088,
"max_allocated": 736.603648
},
"latency": {
"unit": "s",
"values": [
0.09656934356689453
],
"count": 1,
"total": 0.09656934356689453,
"mean": 0.09656934356689453,
"p50": 0.09656934356689453,
"p90": 0.09656934356689453,
"p95": 0.09656934356689453,
"p99": 0.09656934356689453,
"stdev": 0,
"stdev_": 0
},
"throughput": null,
"energy": {
"unit": "kWh",
"cpu": 9.655172959444339e-05,
"ram": 5.289819798409629e-05,
"gpu": 0.00012885288085999827,
"total": 0.000278302808438538
},
"efficiency": null
},
"forward": {
"memory": {
"unit": "MB",
"max_ram": 1261.244416,
"max_global_vram": 1444.41344,
"max_process_vram": 0.0,
"max_reserved": 796.91776,
"max_allocated": 757.115392
},
"latency": {
"unit": "s",
"values": [
0.12223385620117187,
0.01460223960876465,
0.014498815536499024,
0.014973952293395995,
0.015240192413330078,
0.014746623992919922,
0.014928895950317383,
0.01489408016204834,
0.014621696472167968,
0.015159296035766602,
0.01496678352355957,
0.014810111999511719,
0.01443222427368164,
0.014828384399414063,
0.015224831581115723,
0.014653471946716309,
0.014630911827087402,
0.014589952468872071,
0.015396863937377929,
0.014947327613830566,
0.014881759643554688,
0.014642175674438476,
0.014335935592651367,
0.014441439628601074,
0.014435327529907227,
0.014470144271850586,
0.013916159629821777,
0.01480294418334961,
0.01449062442779541,
0.014253055572509766,
0.014006272315979004,
0.014315520286560059,
0.014427136421203614,
0.014754816055297852,
0.014408703804016113,
0.014225407600402832,
0.013711296081542968,
0.013991968154907226,
0.01420076847076416,
0.014361599922180175,
0.01420582389831543,
0.014324799537658691,
0.014434304237365723,
0.014057472229003906,
0.014395392417907715,
0.015082528114318847,
0.014482432365417481,
0.015108096122741698,
0.015624192237854004,
0.015576064109802246,
0.014568448066711426,
0.014230527877807618,
0.014285823822021485,
0.014238752365112305,
0.014342144012451171,
0.013947903633117676,
0.014435327529907227,
0.014358528137207031,
0.013988863945007325,
0.014299136161804199,
0.014434304237365723,
0.014012415885925293
],
"count": 62,
"total": 1.009886848449707,
"mean": 0.01628849755564044,
"p50": 0.014455791950225831,
"p90": 0.015154176044464111,
"p95": 0.015389030361175535,
"p99": 0.05720196118354803,
"stdev": 0.013570939510151937,
"stdev_": 83.31609139390848
},
"throughput": {
"unit": "samples/s",
"value": 122.78603309900936
},
"energy": {
"unit": "kWh",
"cpu": 1.7260016728067008e-07,
"ram": 9.421039191321508e-08,
"gpu": 4.0845443304348045e-07,
"total": 6.752649922373661e-07
},
"efficiency": {
"unit": "samples/kWh",
"value": 2961800.216198634
}
}
}
}