Closed kurdt23 closed 6 months ago
The released config is what we used for the paper. I'd expect small fluctuations if you re-train the model, should not change the overall conclusion.
I'll release our ckpt soon, you may compare your results with it too.
Closing, feel free to re-open if further discussion is needed.
Hello, I would like to thank you for your excellent work. I read your article and ran the model at 10 epochs, but my metric is a little different from the one presented in the article. I could not find the exact number of epochs in the article. Please can you tell me what is the number of epochs you are using in the training in the article?