Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

the traing log like this is Normal? I do not find loss in the logs, and what does the "grad norm: nan" mean? #396

Open
alphanlp opened this issue Aug 27, 2023 · 0 comments

Comments

@alphanlp
Copy link

d norm: nan | actual seqlen: 2048 | number of skipped iterations: 0 | number of nan iterations: 0 | samples per second: 1.886 | TFLOPs: 78.46 |
iteration 5426/ 250000 | consumed samples: 43408 | consumed tokens: 88899584 | elapsed time per iteration (ms): 4247.9 | learning rate: 2.999E-04 | global batch size: 8 | loss scale: 1.0 | grad norm: nan | actual seqlen: 2048 | number of skipped iterations: 0 | number of nan iterations: 0 | samples per second: 1.883 | TFLOPs: 78.36 |
iteration 5427/ 250000 | consumed samples: 43416 | consumed tokens: 88915968 | elapsed time per iteration (ms): 4225.8 | learning rate: 2.999E-04 | global batch size: 8 | loss scale: 1.0 | grad norm: nan | actual seqlen: 2048 | number of skipped iterations: 0 | number of nan iterations: 0 | samples per second: 1.893 | TFLOPs: 78.77 |
iteration 5428/ 250000 | consumed samples: 43424 | consumed tokens: 88932352 | elapsed time per iteration (ms): 4229.2 | learning rate: 2.999E-04 | global batch size: 8 | loss scale: 1.0 | grad norm: nan | actual seqlen: 2048 | number of skipped iterations: 0 | number of nan iterations: 0 | samples per second: 1.892 | TFLOPs: 78.71 |
iteration 5429/ 250000 | consumed samples: 43432 | consumed tokens: 88948736 | elapsed time per iteration (ms): 4233.6 | learning rate: 2.999E-04 | global batch size: 8 | loss scale: 1.0 | grad norm: nan | actual seqlen: 2048 | number of skipped iterations: 0 | number of nan iterations: 0 | samples per second: 1.890 | TFLOPs: 78.63 |
iteration 5430/ 250000 | consumed samples: 43440 | consumed tokens: 88965120 | elapsed time per iteration (ms): 4247.0 | learning rate: 2.999E-04 | global batch size: 8 | loss scale: 1.0 | grad norm: nan | actual seqlen: 2048 | number of skipped iterations: 0 | number of nan iterations: 0 | samples per second: 1.884 | TFLOPs: 78.38 |

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant