IlyasMoutawwakil's picture
Upload cuda_inference_transformers_text-generation_openai-community/gpt2/benchmark.json with huggingface_hub
b710af5 verified
raw
history blame
13.9 kB
{
"config": {
"name": "cuda_inference_transformers_text-generation_openai-community/gpt2",
"backend": {
"name": "pytorch",
"version": "2.3.0+cu121",
"_target_": "optimum_benchmark.backends.pytorch.backend.PyTorchBackend",
"task": "text-generation",
"library": "transformers",
"model": "openai-community/gpt2",
"processor": "openai-community/gpt2",
"device": "cuda",
"device_ids": "0",
"seed": 42,
"inter_op_num_threads": null,
"intra_op_num_threads": null,
"model_kwargs": {},
"processor_kwargs": {},
"hub_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": 1,
"num_choices": 2,
"sequence_length": 2
},
"new_tokens": null,
"latency": true,
"memory": 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",
"start_method": "spawn"
},
"environment": {
"cpu": " AMD EPYC 7R32",
"cpu_count": 16,
"cpu_ram_mb": 66697.29792,
"system": "Linux",
"machine": "x86_64",
"platform": "Linux-5.10.215-203.850.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": "0.2.1",
"optimum_benchmark_commit": null,
"transformers_version": "4.40.2",
"transformers_commit": null,
"accelerate_version": "0.30.1",
"accelerate_commit": null,
"diffusers_version": "0.27.2",
"diffusers_commit": null,
"optimum_version": null,
"optimum_commit": null,
"timm_version": "1.0.3",
"timm_commit": null,
"peft_version": null,
"peft_commit": null
}
},
"report": {
"prefill": {
"memory": {
"unit": "MB",
"max_ram": 1070.706688,
"max_global_vram": 1365.77024,
"max_process_vram": 0.0,
"max_reserved": 725.614592,
"max_allocated": 522.007552
},
"latency": {
"unit": "s",
"count": 57,
"total": 0.49521626186370843,
"mean": 0.008688004594100148,
"stdev": 0.0003397676374341216,
"p50": 0.008515168190002441,
"p90": 0.0089687105178833,
"p95": 0.00946518383026123,
"p99": 0.00983262493133545,
"values": [
0.00995686435699463,
0.009408991813659668,
0.009735008239746094,
0.009059455871582031,
0.00886787223815918,
0.008844256401062012,
0.008893407821655274,
0.00839635181427002,
0.008380831718444825,
0.008586432456970215,
0.00968995189666748,
0.009003520011901855,
0.008945504188537597,
0.008483327865600587,
0.008500672340393067,
0.008490719795227052,
0.008476256370544433,
0.008463552474975586,
0.00834335994720459,
0.008435775756835937,
0.008485728263854981,
0.008657504081726074,
0.008507967948913574,
0.008510720252990723,
0.00842959976196289,
0.008433055877685548,
0.008515168190002441,
0.008593567848205566,
0.008474752426147461,
0.008499168395996093,
0.008600895881652831,
0.00854150390625,
0.008482912063598632,
0.008535584449768067,
0.008491583824157715,
0.008490912437438965,
0.0086561279296875,
0.008509119987487794,
0.008736991882324218,
0.008874208450317382,
0.008841216087341308,
0.008907008171081542,
0.008904095649719238,
0.00888912010192871,
0.008821951866149902,
0.008400128364562989,
0.008374848365783692,
0.008483648300170898,
0.008844927787780761,
0.008819392204284668,
0.008851136207580566,
0.008826656341552734,
0.008462464332580567,
0.008438464164733886,
0.008440032005310058,
0.008473600387573242,
0.008448384284973144
]
},
"throughput": {
"unit": "tokens/s",
"value": 230.20245654084488
},
"energy": {
"unit": "kWh",
"cpu": 1.0866128310758648e-07,
"ram": 5.9411654925964975e-08,
"gpu": 1.8366222644037274e-07,
"total": 3.5173516447392423e-07
},
"efficiency": {
"unit": "tokens/kWh",
"value": 5686096.250829278
}
},
"decode": {
"memory": {
"unit": "MB",
"max_ram": 1070.706688,
"max_global_vram": 1365.77024,
"max_process_vram": 0.0,
"max_reserved": 725.614592,
"max_allocated": 522.4576
},
"latency": {
"unit": "s",
"count": 57,
"total": 0.5130666542053224,
"mean": 0.009001169372023199,
"stdev": 0.00036550224354141097,
"p50": 0.008815936088562012,
"p90": 0.009341004562377929,
"p95": 0.00987852783203125,
"p99": 0.010171902770996093,
"values": [
0.010046815872192383,
0.009864959716796875,
0.010331104278564453,
0.009186400413513183,
0.00917407989501953,
0.0091843843460083,
0.009119839668273925,
0.008674880027770997,
0.008725472450256348,
0.00993280029296875,
0.009528800010681153,
0.009269791603088379,
0.009339263916015625,
0.008746111869812011,
0.008746015548706054,
0.00876636791229248,
0.008769920349121094,
0.008798815727233887,
0.008631296157836914,
0.008736767768859864,
0.008719936370849609,
0.008782784461975097,
0.008815936088562012,
0.008767583847045898,
0.008659071922302246,
0.008852671623229981,
0.008956416130065918,
0.00872873592376709,
0.008797151565551758,
0.00881817626953125,
0.008794528007507324,
0.008790143966674805,
0.008812095642089844,
0.008768383979797362,
0.008778944015502929,
0.008898591995239258,
0.008841631889343261,
0.008771552085876464,
0.009266207695007325,
0.009234975814819336,
0.009343615531921387,
0.00920751953125,
0.009206015586853028,
0.00913593578338623,
0.009159808158874511,
0.008656895637512207,
0.008671104431152344,
0.008721599578857422,
0.00921737575531006,
0.009160832405090332,
0.009150752067565918,
0.009152864456176758,
0.008734368324279785,
0.008721792221069335,
0.008855487823486328,
0.008798687934875489,
0.008742591857910157
]
},
"throughput": {
"unit": "tokens/s",
"value": 111.09667629498558
},
"energy": {
"unit": "kWh",
"cpu": 1.0476376458805027e-07,
"ram": 5.70648576330448e-08,
"gpu": 1.7437873459533498e-07,
"total": 3.362073568164301e-07
},
"efficiency": {
"unit": "tokens/kWh",
"value": 2974354.9024895434
}
},
"per_token": {
"memory": null,
"latency": {
"unit": "s",
"count": 56,
"total": 0.9899519977569576,
"mean": 0.017677714245659965,
"stdev": 0.0006215098263446906,
"p50": 0.017354239463806154,
"p90": 0.01831676769256592,
"p95": 0.01862886333465576,
"p99": 0.01970159015655518,
"values": [
0.019314687728881837,
0.020143104553222657,
0.01839206314086914,
0.01807257652282715,
0.01804902458190918,
0.018050048828125,
0.017108991622924806,
0.017132543563842775,
0.01840025520324707,
0.01934035110473633,
0.01833465576171875,
0.018298879623413086,
0.017289215087890625,
0.017254400253295898,
0.017297407150268555,
0.017274879455566407,
0.01726464080810547,
0.01703424072265625,
0.017189920425415037,
0.017233888626098634,
0.017458175659179686,
0.01735577583312988,
0.017310720443725586,
0.017108991622924806,
0.017306623458862306,
0.017500160217285156,
0.017352703094482422,
0.017292287826538084,
0.017336320877075196,
0.017426431655883787,
0.017366016387939453,
0.017316864013671874,
0.017332223892211913,
0.017294336318969726,
0.017358879089355468,
0.017581024169921876,
0.017306623458862306,
0.018010112762451173,
0.018136064529418947,
0.01820467185974121,
0.018152448654174806,
0.018134016036987305,
0.018053119659423827,
0.017999872207641602,
0.017113088607788086,
0.01706188774108887,
0.01722777557373047,
0.01807360076904297,
0.01800704002380371,
0.01802956771850586,
0.01801113510131836,
0.0172359676361084,
0.01719398307800293,
0.017314815521240236,
0.01728102493286133,
0.01723187255859375
]
},
"throughput": {
"unit": "tokens/s",
"value": 56.56839940409767
},
"energy": null,
"efficiency": null
}
}
}