Training loss and acc/auc curve

[wwx13]

Hi, I'm using amos now. My amos model (small size, discriminator ) have a low recall (70-80% percision while 40% recall). 60% mlm acc of generator. I would just like to ask if you can post the loss of both base and large models (or even share the loss training curve, acc curve or auc curve ) so that i have a kind of reference point when training own models. This will help me a lot!

Thank u.

[yumeng5]

Hi @wwx13

We only have the logs for base models. The logs contain the following metrics which I believe are relevant to your question.

• Discriminator accuracy on all tokens:

  

• Discriminator accuracy on non-replaced tokens:

  

• Discriminator accuracy on replaced tokens:

  

• Generator token replace rate (defined to be the portion of replaced tokens by the generator out of all tokens):

  

I hope these are helpful. Let me know if you have further questions.

Thanks, Yu