ferdinand.mom
add metrics + profiler.csv for 16 GPUS
b79f60a
raw
history blame
2.1 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
1,4190000.0000000005,105000.0,40000.0,2500.0,1020.0,11.1,0.0001,22.7,22.7,25.6,3274.04,3274.05,13040.0
2,8390000.0,50800.0,82500.0,5160.0,1020.0,11.1,9.53e-05,46.8,46.8,25.9,3274.04,12603.56,13042.0
3,12600000.0,50900.0,82300.0,5150.0,1020.0,9.9,9.05e-05,46.7,46.7,40.4,3274.04,3274.05,13042.0
4,16800000.0,47800.0,87700.0,5480.0,1020.0,11.9,8.58e-05,49.7,49.7,61.2,3274.04,12603.56,13042.0
5,21000000.0,46600.0,90100.0,5630.0,1020.0,9.05,8.11e-05,51.1,51.1,8.31,,,
6,25200000.0,51500.0,81500.0,5090.0,1020.0,8.86,7.63e-05,46.2,46.2,6.63,3274.04,12603.56,13042.0
7,29400000.0,534000.0,7850.0,491.0,1020.0,8.37,7.16e-05,4.45,4.45,4.93,3274.04,12603.56,13042.0
8,33600000.0,53800.0,78000.0,4880.0,1020.0,7.97,6.68e-05,44.2,44.2,3.13,3274.04,12603.56,13042.0
9,37700000.0,47500.0,88400.0,5520.0,1020.0,7.83,6.21e-05,50.1,50.1,9.04,3274.04,12603.56,13042.0
10,41900000.0,48000.0,87300.0,5460.0,1020.0,7.62,5.74e-05,49.5,49.5,5.09,3274.04,12603.56,13042.0
11,46100000.0,47900.0,87600.0,5470.0,1020.0,7.47,5.26e-05,49.7,49.7,4.06,,,
12,50300000.0,45700.0,91700.0,5730.0,1020.0,7.34,4.79e-05,52.0,52.0,3.12,3274.04,12603.56,13042.0
13,54500000.0,48000.0,87400.0,5460.0,1020.0,7.23,4.32e-05,49.6,49.6,2.73,3274.04,12603.56,13042.0
14,58700000.0,44700.0,93800.0,5860.0,1020.0,7.14,3.84e-05,53.2,53.2,2.33,3274.04,12603.56,13042.0
15,62900000.0,46200.0,90800.0,5680.0,1020.0,7.06,3.37e-05,51.5,51.5,2.48,3274.04,12603.56,13042.0
16,67099999.99999999,47000.0,89200.0,5580.0,1020.0,6.98,2.89e-05,50.6,50.6,2.66,3274.04,12603.56,13042.0
17,71300000.0,46000.0,91100.0,5690.0,1020.0,6.9,2.42e-05,51.7,51.7,1.89,,,
18,75500000.0,43500.0,96400.0,6030.0,1020.0,6.84,1.95e-05,54.7,54.7,1.61,3274.04,12603.56,13042.0
19,79700000.0,47500.0,88200.0,5510.0,1020.0,6.8,1.47e-05,50.0,50.0,1.85,,,
20,83900000.0,44700.0,93900.0,5870.0,1020.0,6.76,1e-05,53.2,53.2,1.81,,,