Kurt Shuster
Kurt Shuster
have you set up your heroku credentials?
Maybe CC @JackUrb who might know why heroku is not working here. Have you searched elsewhere for this issue? It may not be related to ParlAI specifically?
Yes, feel free to come back and update us when you find something!
Thank you for your feedback and really appreciate your help in improving the bot! We are indeed aware of these issues and look forward to improving them in future iterations...
Thanks for flagging, the fix is up now: #4786
The BB3 3B model is a bit more memory intensive than BB2. In my local experiments I can get it to train on 2 x 32gb GPUs, with `--text-truncate 128...
Only for training. Inference could easily be run on 1 40gb node (I can run it on 1 16gpb node)
That's... weird. You used the exact same command as above?
Looks like there's some issues with sharding on the GPUs? I was able to run your exact command with 2 x 32gb v100s, not sure what's going on here. Note...
you can try just training a vanilla dialogue model, assuming there exists a KG dialogue dataset in ParlAI. If there doesn't, first step is to add one!