Closed kengkeng1 closed 1 year ago
Hi @kengkeng1,
Sorry, we have not saved the pretrain logs as it was trained on some temporary servers. However, since it may be a help to you, we will rerun the pertaining process and provide a log for reference recently when we have time.
Thanks for your kind attention! I'm so sorry for the late reply. We retrain the model and upload the log file into google drive (https://drive.google.com/file/d/1Ttk8b6uaOOOXYVAd6tDr8fQeqTuFLIrn/view?usp=drive_link). You can test the 250/275/300 epoch checkpoints for the best performance.
Thanks for your amazing work! Can you provide the pretain logs file.I want to check if I'm running it incorrectly. Thank you!