PII_Bert / real_test_inference_metrics_T4.csv
PuxAI's picture
Add real Test Dataset inference benchmarks from T4 GPU
84f5008 verified
model,gpu_used,test_samples,total_valid_tokens,total_time_seconds,latency_ms_per_sample,throughput_samples_per_sec,throughput_tokens_per_sec
bert-base-cased,CPU,2176,111284,254.09,116.77,8.56,437.97
roberta-base,CPU,2176,102023,253.81,116.64,8.57,401.97
distilbert-base-cased,CPU,2176,111284,130.62,60.03,16.66,851.96
albert-base-v2,CPU,2176,103998,310.02,142.47,7.02,335.45
FacebookAI-xlm-roberta-base,CPU,2176,106333,255.48,117.41,8.52,416.21
google-bert-bert-base-multilingual-cased,CPU,2176,111171,253.72,116.6,8.58,438.17