new data

logs/main_log.txt

@@ -55315,3 +55315,15 @@ time (ms)
 time (ms)
  iteration     5870/  159576 | consumed samples:       170928 | elapsed time per iteration (ms): 16844.9 | learning rate: 4.727E-05 | global batch size:    64 | lm loss: 6.372821E+00 | loss scale: 2048.0 | grad norm: 49107.159 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
 time (ms)
+ iteration     5880/  159576 | consumed samples:       171568 | elapsed time per iteration (ms): 16812.2 | learning rate: 4.745E-05 | global batch size:    64 | lm loss: 6.379050E+00 | loss scale: 2048.0 | grad norm: 76898.126 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
+time (ms)
+ iteration     5890/  159576 | consumed samples:       172208 | elapsed time per iteration (ms): 16819.7 | learning rate: 4.763E-05 | global batch size:    64 | lm loss: 6.333071E+00 | loss scale: 2048.0 | grad norm: 69874.656 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
+time (ms)
+ iteration     5900/  159576 | consumed samples:       172848 | elapsed time per iteration (ms): 16821.3 | learning rate: 4.780E-05 | global batch size:    64 | lm loss: 6.354385E+00 | loss scale: 2048.0 | grad norm: 57915.999 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
+time (ms)
+ iteration     5910/  159576 | consumed samples:       173488 | elapsed time per iteration (ms): 16679.9 | learning rate: 4.798E-05 | global batch size:    64 | lm loss: 6.361916E+00 | loss scale: 2048.0 | grad norm: 56535.869 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
+time (ms)
+ iteration     5920/  159576 | consumed samples:       174128 | elapsed time per iteration (ms): 16731.8 | learning rate: 4.816E-05 | global batch size:    64 | lm loss: 6.371978E+00 | loss scale: 2048.0 | grad norm: 75613.913 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
+time (ms)
+ iteration     5930/  159576 | consumed samples:       174768 | elapsed time per iteration (ms): 16796.3 | learning rate: 4.834E-05 | global batch size:    64 | lm loss: 6.373956E+00 | loss scale: 2048.0 | grad norm: 64436.905 | num zeros: 0.0 | number of skipped iterations:   0 | number of nan iterations:   0 |
+time (ms)