GPU Benchmark result

GPU Memory & Inference Time check according to sequence length

Environment Info

transformers_version: 4.11.3
framework: PyTorch
use_torchscript: False
framework_version: 1.7.1
python_version: 3.6.13
system: Linux
cpu: x86_64
architecture: 64bit
fp16: False
use_multiprocessing: True
only_pretrain_model: False
cpu_ram_mb: 15717
use_gpu: True
num_gpus: 1
gpu: Tesla T4
gpu_ram_mb: 15109
gpu_power_watts: 70.0
gpu_performance_state: 0
use_tpu: False

Benchmark result