ferdinand.mom
add metrics + profiler.csv for 16 GPUS
b79f60a
raw
history blame
2.14 kB
iteration,consumed_tokens,elapsed_time_per_iteration_ms,tokens_per_sec,tokens_per_sec_per_gpu,global_batch_size,lm_loss,lr,model_tflops_per_gpu,hardware_tflops_per_gpu,grad_norm,memory_usage_MiB,peak_allocated_MiB,peak_reserved_MiB
1,4190000.0000000005,63000.0,66600.0,4160.0,1020.0,11.1,0.0001,37.8,37.8,24.9,3168.18,4459.05,23598.0
2,8390000.0,32100.0,131000.0,8170.0,1020.0,11.1,9.53e-05,74.1,74.1,25.1,3168.18,4459.05,23598.0
3,12600000.0,29300.0,143000.0,8950.0,1020.0,9.49,9.05e-05,81.2,81.2,21.5,3168.18,4459.05,23598.0
4,16800000.0,28900.0,145000.0,9070.0,1020.0,9.36,8.58e-05,82.3,82.3,21.4,3168.18,22834.67,23598.0
5,21000000.0,31800.0,132000.0,8240.0,1020.0,9.01,8.11e-05,74.8,74.8,12.7,,,
6,25200000.0,30800.0,136000.0,8520.0,1020.0,10.3,7.63e-05,77.3,77.3,47.1,3168.18,22834.67,23598.0
7,29400000.0,241000.0,17400.0,1090.0,1020.0,8.68,7.16e-05,9.88,9.88,5.58,3168.18,22834.67,23598.0
8,33600000.0,31200.0,135000.0,8410.0,1020.0,8.32,6.68e-05,76.3,76.3,4.77,3168.18,22834.67,23598.0
9,37700000.0,31900.0,131000.0,8210.0,1020.0,7.95,6.21e-05,74.5,74.5,3.31,3168.18,22834.67,23598.0
10,41900000.0,30600.0,137000.0,8550.0,1020.0,7.69,5.74e-05,77.6,77.6,4.31,3168.18,22834.67,23598.0
11,46100000.0,32100.0,131000.0,8170.0,1020.0,7.45,5.26e-05,74.2,74.2,2.5,,,
12,50300000.0,31700.0,132000.0,8270.0,1020.0,7.37,4.79e-05,75.0,75.0,5.02,3168.18,22834.67,23598.0
13,54500000.0,32200.000000000004,130000.0,8150.0,1020.0,7.31,4.32e-05,73.9,73.9,6.06,3168.18,22834.67,23598.0
14,58700000.0,31900.0,132000.0,8220.0,1020.0,7.19,3.84e-05,74.6,74.6,5.3,,,
15,62900000.0,34100.0,123000.0,7690.0,1020.0,7.06,3.37e-05,69.7,69.7,2.73,3168.18,22834.67,23598.0
16,67099999.99999999,32500.0,129000.0,8060.000000000001,1020.0,6.97,2.89e-05,73.2,73.2,1.99,3168.18,22834.67,23598.0
17,71300000.0,30800.0,136000.0,8510.0,1020.0,6.91,2.42e-05,77.2,77.2,2.04,3168.18,22834.67,23598.0
18,75500000.0,32900.0,128000.0,7970.0,1020.0,6.86,1.95e-05,72.4,72.4,2.0,3168.18,22834.67,23598.0
19,79700000.0,30100.0,139000.0,8700.0,1020.0,6.81,1.47e-05,78.9,78.9,2.01,,,
20,83900000.0,31100.0,135000.0,8420.0,1020.0,6.77,1e-05,76.4,76.4,1.94,,,