Skip to content

high train loss #130

@eleveneee

Description

@eleveneee

First of all, thank you for your excellent work!
I have not modified any other parameters, but my losses are still high after 25 rounds of training. Do I still need to continue training? Is this normal?
Look forward to your reply!
Here is the specific situation:

23100.7s 1674 Mon Mar 27 23:05:44 2023 Epoch: 24 Sample 7120/7200 Loss: 0.4667869508266449
23104.6s 1675 Mon Mar 27 23:05:48 2023 Epoch: 24 Sample 7130/7200 Loss: 0.5839558243751526
23108.4s 1676 Mon Mar 27 23:05:52 2023 Epoch: 24 Sample 7140/7200 Loss: 0.4183468520641327
23112.2s 1677 Mon Mar 27 23:05:55 2023 Epoch: 24 Sample 7150/7200 Loss: 0.6217041611671448
23116.2s 1678 Mon Mar 27 23:05:59 2023 Epoch: 24 Sample 7160/7200 Loss: 0.5632153153419495
23120.8s 1679 Mon Mar 27 23:06:03 2023 Epoch: 24 Sample 7170/7200 Loss: 0.47710853815078735
23123.9s 1680 Mon Mar 27 23:06:07 2023 Epoch: 24 Sample 7180/7200 Loss: 0.6275425553321838
23127.7s 1681 Mon Mar 27 23:06:11 2023 Epoch: 24 Sample 7190/7200 Loss: 0.42571431398391724
23143.9s 1682 Current learning rate> 0.0001
23143.9s 1683 -------------------------------------------------------
23143.9s 1684 DexiNed, # of Parameters:
23143.9s 1685 35215245

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions