ferdinand.mom
add metrics + profiler.csv for 16 GPUS
b79f60a
raw
history blame
2.22 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
1,4190000.0000000005,36400.0,115000.0,7200.0,1020.0,11.2,0.0001,65.3,65.3,12.1,1572.71,31525.69,31874.0
2,8390000.0,18000.0,233000.0,14500.0,1020.0,11.2,9.53e-05,132.0,132.0,12.2,1572.71,31525.69,31874.0
3,12600000.0,18100.0,232000.0,14500.0,1020.0,10.0,9.05e-05,132.0,132.0,51.6,1572.71,31525.69,31874.0
4,16800000.0,17400.0,241000.0,15000.0,1020.0,11.7,8.58e-05,136.0,136.0,18.2,1572.71,1572.75,31874.0
5,21000000.0,17500.0,239000.0,14900.0,1020.0,10.4,8.11e-05,136.0,136.0,16.0,1572.71,31525.69,31874.0
6,25200000.0,17500.0,240000.0,15000.0,1020.0,9.9,7.63e-05,136.0,136.0,9.07,1572.71,31525.69,31874.0
7,29400000.0,216000.0,19400.0,1210.0,1020.0,9.37,7.16e-05,11.0,11.0,6.23,1572.71,31525.69,31874.0
8,33600000.0,17400.0,241000.0,15100.0,1020.0,8.89,6.68e-05,137.0,137.0,5.76,1572.71,31525.69,31874.0
9,37700000.0,18300.0,229000.0,14300.0,1020.0,8.8,6.21e-05,130.0,130.0,11.2,,,
10,41900000.0,17300.0,243000.0,15200.0,1020.0,8.33,5.74e-05,138.0,138.0,5.72,1572.71,31525.69,31874.0
11,46100000.0,17200.0,243000.0,15200.0,1020.0,8.06,5.26e-05,138.0,138.0,4.91,1572.71,31525.69,31874.0
12,50300000.0,17000.0,247000.0,15400.0,1020.0,7.9,4.79e-05,140.0,140.0,4.86,1572.71,31525.69,31874.0
13,54500000.0,17300.0,242000.0,15100.0,1020.0,7.75,4.32e-05,137.0,137.0,4.69,,,
14,58700000.0,17300.0,243000.0,15200.0,1020.0,7.62,3.84e-05,138.0,138.0,4.69,1572.71,31525.69,31874.0
15,62900000.0,17300.0,242000.0,15100.0,1020.0,7.48,3.37e-05,137.0,137.0,4.49,1572.71,31525.69,31874.0
16,67099999.99999999,17000.0,247000.0,15400.0,1020.0,7.34,2.89e-05,140.0,140.0,3.99,1572.71,31525.69,31874.0
17,71300000.0,17100.0,245000.0,15300.0,1020.0,7.23,2.42e-05,139.0,139.0,3.54,1572.71,31525.69,31874.0
18,75500000.0,17300.0,242000.0,15100.0,1020.0,7.16,1.95e-05,137.0,137.0,3.28,1572.71,31525.69,31874.0
19,79700000.0,17300.0,242000.0,15100.0,1020.0,7.09,1.47e-05,137.0,137.0,3.2,1572.71,31525.69,31874.0
20,83900000.0,17500.0,240000.0,15000.0,1020.0,7.03,1e-05,136.0,136.0,3.1,,,