-
Notifications
You must be signed in to change notification settings - Fork 731
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Create a table of results for our supported checkpoints #142
Labels
documentation
Improvements or additions to documentation
Comments
that's a great idea. Maybe also adding a Num GPU column since some models can't be run on a single device. (Once the current QLoRA experiments are done and datasets are added, I can maybe get started on this) |
is this resolved or there is more to be done? |
We should probably also have a similar table for finetuning. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
We support a large number of checkpoints. And there's a multitude of scripts that can be run.
Users often ask questions like "can I run X script with Y model given Z memory?" or "is X (script, model) faster than Y (script, model)?"
The idea would be to collect data in a Markdown table that we can point to answer these questions.
The data should always be collected from the same machine (our 8xA100 node).
Some scripts will have to specify the hparams used.
We can pick out a subset of the checkpoints to start with.
For example:
generate/base.py --precision bf16-true
The text was updated successfully, but these errors were encountered: