-
-
Save pzread/2be3b2db7c0ffa14518085f08e33b815 to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
{ | |
"trigger": { | |
"timestamp": "1688166610" | |
}, | |
"benchmarks": [ | |
{ | |
"definition": { | |
"benchmark_id": "173c7180-bad4-4b91-8423-4beeb13d2b0a-MODEL_T5_LARGE-fp32-JAX-512xi32-batch1", | |
"benchmark_name": "T5_LARGE_FP32_JAX_512XI32_BATCH1", | |
"framework": "ModelFrameworkType.JAX", | |
"data_type": "DataType.FP32", | |
"batch_size": 1, | |
"inputs": [ | |
"1x512xi32", | |
"1x512xi32" | |
], | |
"outputs": [ | |
"1x512x1024xf32" | |
], | |
"compiler": "xla", | |
"device": "gpu", | |
"tags": [ | |
"fp32", | |
"transformer-encoder", | |
"transformer-decoder", | |
"t5", | |
"batch-1" | |
] | |
}, | |
"metrics": { | |
"framework_level": { | |
"min_warmup_latency_ms": 20.655435044318438, | |
"max_warmup_latency_ms": 50176.288513001055, | |
"mean_warmup_latency_ms": 10054.563094791956, | |
"median_warmup_latency_ms": 25.850410922430456, | |
"stddev_warmup_latency_ms": 22428.7264534675, | |
"warmup_iterations": 5, | |
"min_latency_ms": 20.552234025672078, | |
"max_latency_ms": 20.809899899177253, | |
"mean_latency_ms": 20.659179559443146, | |
"median_latency_ms": 20.647069963160902, | |
"stddev_latency_ms": 0.05457004473826859, | |
"benchmark_iterations": 50, | |
"compile_time_s": 50.176288513001055, | |
"input_data_transfer_ms": 0.5347649566829205 | |
} | |
} | |
}, | |
{ | |
"definition": { | |
"benchmark_id": "173c7180-bad4-4b91-8423-4beeb13d2b0a-MODEL_T5_LARGE-fp32-JAX-512xi32-batch16", | |
"benchmark_name": "T5_LARGE_FP32_JAX_512XI32_BATCH16", | |
"framework": "ModelFrameworkType.JAX", | |
"data_type": "DataType.FP32", | |
"batch_size": 16, | |
"inputs": [ | |
"16x512xi32", | |
"16x512xi32" | |
], | |
"outputs": [ | |
"16x512x1024xf32" | |
], | |
"compiler": "xla", | |
"device": "gpu", | |
"tags": [ | |
"fp32", | |
"transformer-encoder", | |
"transformer-decoder", | |
"t5", | |
"batch-16" | |
] | |
}, | |
"metrics": { | |
"framework_level": { | |
"min_warmup_latency_ms": 245.36785506643355, | |
"max_warmup_latency_ms": 44760.09083108511, | |
"mean_warmup_latency_ms": 9148.438019421883, | |
"median_warmup_latency_ms": 245.4743740381673, | |
"stddev_warmup_latency_ms": 19907.51912043897, | |
"warmup_iterations": 5, | |
"min_latency_ms": 245.34180504269898, | |
"max_latency_ms": 245.61262398492545, | |
"mean_latency_ms": 245.50913302926347, | |
"median_latency_ms": 245.5175855429843, | |
"stddev_latency_ms": 0.04777120683995344, | |
"benchmark_iterations": 50, | |
"compile_time_s": 44.76009083108511, | |
"input_data_transfer_ms": 0.4104010295122862 | |
} | |
} | |
}, | |
{ | |
"definition": { | |
"benchmark_id": "173c7180-bad4-4b91-8423-4beeb13d2b0a-MODEL_T5_LARGE-fp32-JAX-512xi32-batch24", | |
"benchmark_name": "T5_LARGE_FP32_JAX_512XI32_BATCH24", | |
"framework": "ModelFrameworkType.JAX", | |
"data_type": "DataType.FP32", | |
"batch_size": 24, | |
"inputs": [ | |
"24x512xi32", | |
"24x512xi32" | |
], | |
"outputs": [ | |
"24x512x1024xf32" | |
], | |
"compiler": "xla", | |
"device": "gpu", | |
"tags": [ | |
"fp32", | |
"transformer-encoder", | |
"transformer-decoder", | |
"t5", | |
"batch-24" | |
] | |
}, | |
"metrics": { | |
"framework_level": { | |
"min_warmup_latency_ms": 368.98510495666414, | |
"max_warmup_latency_ms": 45109.30842696689, | |
"mean_warmup_latency_ms": 9317.142666201107, | |
"median_warmup_latency_ms": 369.1369229927659, | |
"stddev_warmup_latency_ms": 20008.42892586601, | |
"warmup_iterations": 5, | |
"min_latency_ms": 368.92264301422983, | |
"max_latency_ms": 369.16676396504045, | |
"mean_latency_ms": 369.0529297059402, | |
"median_latency_ms": 369.05131745152175, | |
"stddev_latency_ms": 0.04959245737189531, | |
"benchmark_iterations": 50, | |
"compile_time_s": 45.10930842696689, | |
"input_data_transfer_ms": 0.5262399790808558 | |
} | |
} | |
}, | |
{ | |
"definition": { | |
"benchmark_id": "173c7180-bad4-4b91-8423-4beeb13d2b0a-MODEL_T5_LARGE-fp32-JAX-512xi32-batch32", | |
"benchmark_name": "T5_LARGE_FP32_JAX_512XI32_BATCH32", | |
"framework": "ModelFrameworkType.JAX", | |
"data_type": "DataType.FP32", | |
"batch_size": 32, | |
"inputs": [ | |
"32x512xi32", | |
"32x512xi32" | |
], | |
"outputs": [ | |
"32x512x1024xf32" | |
], | |
"compiler": "xla", | |
"device": "gpu", | |
"tags": [ | |
"fp32", | |
"transformer-encoder", | |
"transformer-decoder", | |
"t5", | |
"batch-32" | |
] | |
}, | |
"metrics": { | |
"framework_level": { | |
"min_warmup_latency_ms": 480.0509420456365, | |
"max_warmup_latency_ms": 45333.85372895282, | |
"mean_warmup_latency_ms": 9450.862502027303, | |
"median_warmup_latency_ms": 480.13286804780364, | |
"stddev_warmup_latency_ms": 20059.201904894326, | |
"warmup_iterations": 5, | |
"min_latency_ms": 479.95751502458006, | |
"max_latency_ms": 480.5134879425168, | |
"mean_latency_ms": 480.10662383865565, | |
"median_latency_ms": 480.09840404847637, | |
"stddev_latency_ms": 0.08080983067290791, | |
"benchmark_iterations": 50, | |
"compile_time_s": 45.33385372895282, | |
"input_data_transfer_ms": 0.5753289442509413 | |
} | |
} | |
}, | |
{ | |
"definition": { | |
"benchmark_id": "173c7180-bad4-4b91-8423-4beeb13d2b0a-MODEL_T5_LARGE-fp32-JAX-512xi32-batch48", | |
"benchmark_name": "T5_LARGE_FP32_JAX_512XI32_BATCH48", | |
"framework": "ModelFrameworkType.JAX", | |
"data_type": "DataType.FP32", | |
"batch_size": 48, | |
"inputs": [ | |
"48x512xi32", | |
"48x512xi32" | |
], | |
"outputs": [ | |
"48x512x1024xf32" | |
], | |
"compiler": "xla", | |
"device": "gpu", | |
"tags": [ | |
"fp32", | |
"transformer-encoder", | |
"transformer-decoder", | |
"t5", | |
"batch-48" | |
] | |
}, | |
"metrics": { | |
"framework_level": { | |
"min_warmup_latency_ms": 721.3618629612029, | |
"max_warmup_latency_ms": 45887.21120392438, | |
"mean_warmup_latency_ms": 9754.539297986776, | |
"median_warmup_latency_ms": 721.3717340491712, | |
"stddev_warmup_latency_ms": 20198.777647594525, | |
"warmup_iterations": 5, | |
"min_latency_ms": 720.7678849808872, | |
"max_latency_ms": 721.505339955911, | |
"mean_latency_ms": 721.1524451547302, | |
"median_latency_ms": 721.028728003148, | |
"stddev_latency_ms": 0.26223423855135464, | |
"benchmark_iterations": 50, | |
"compile_time_s": 45.88721120392438, | |
"input_data_transfer_ms": 0.5476829828694463 | |
} | |
} | |
}, | |
{ | |
"definition": { | |
"benchmark_id": "173c7180-bad4-4b91-8423-4beeb13d2b0a-MODEL_T5_LARGE-fp32-JAX-512xi32-batch64", | |
"benchmark_name": "T5_LARGE_FP32_JAX_512XI32_BATCH64", | |
"framework": "ModelFrameworkType.JAX", | |
"data_type": "DataType.FP32", | |
"batch_size": 64, | |
"inputs": [ | |
"64x512xi32", | |
"64x512xi32" | |
], | |
"outputs": [ | |
"64x512x1024xf32" | |
], | |
"compiler": "xla", | |
"device": "gpu", | |
"tags": [ | |
"fp32", | |
"transformer-encoder", | |
"transformer-decoder", | |
"t5", | |
"batch-64" | |
] | |
}, | |
"metrics": { | |
"framework_level": {} | |
} | |
}, | |
{ | |
"definition": { | |
"benchmark_id": "173c7180-bad4-4b91-8423-4beeb13d2b0a-MODEL_T5_LARGE-fp32-JAX-512xi32-batch512", | |
"benchmark_name": "T5_LARGE_FP32_JAX_512XI32_BATCH512", | |
"framework": "ModelFrameworkType.JAX", | |
"data_type": "DataType.FP32", | |
"batch_size": 512, | |
"inputs": [ | |
"512x512xi32", | |
"512x512xi32" | |
], | |
"outputs": [ | |
"512x512x1024xf32" | |
], | |
"compiler": "xla", | |
"device": "gpu", | |
"tags": [ | |
"fp32", | |
"transformer-encoder", | |
"transformer-decoder", | |
"t5", | |
"batch-512" | |
] | |
}, | |
"metrics": { | |
"framework_level": {} | |
} | |
} | |
], | |
"execution_environment": { | |
"python_environment": { | |
"absl-py": "1.4.0", | |
"certifi": "2023.5.7", | |
"charset-normalizer": "3.1.0", | |
"chex": "0.1.81", | |
"dm-tree": "0.1.8", | |
"etils": "1.3.0", | |
"filelock": "3.12.2", | |
"flax": "0.6.11", | |
"fsspec": "2023.6.0", | |
"huggingface-hub": "0.15.1", | |
"idna": "3.4", | |
"importlib-resources": "5.12.0", | |
"jax": "0.4.13", | |
"jaxlib": "0.4.13+cuda11.cudnn86", | |
"markdown-it-py": "3.0.0", | |
"mdurl": "0.1.2", | |
"ml-dtypes": "0.2.0", | |
"msgpack": "1.0.5", | |
"nest-asyncio": "1.5.6", | |
"numpy": "1.25.0", | |
"opt-einsum": "3.3.0", | |
"optax": "0.1.5", | |
"orbax-checkpoint": "0.2.7", | |
"packaging": "23.1", | |
"Pillow": "9.5.0", | |
"pip": "23.1.2", | |
"protobuf": "4.23.3", | |
"Pygments": "2.15.1", | |
"PyYAML": "6.0", | |
"regex": "2023.6.3", | |
"requests": "2.31.0", | |
"rich": "13.4.2", | |
"safetensors": "0.3.1", | |
"scipy": "1.11.1", | |
"setuptools": "59.6.0", | |
"tensorstore": "0.1.39", | |
"tokenizers": "0.13.3", | |
"toolz": "0.12.0", | |
"tqdm": "4.65.0", | |
"transformers": "4.30.2", | |
"typing_extensions": "4.7.0", | |
"urllib3": "2.0.3", | |
"zipp": "3.15.0" | |
} | |
} | |
} |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment