Skip to content

Allow specifying datasets in configs, change default optimizer, improved CUDA support#1

Merged
aronson merged 6 commits intomainfrom
fizzAI-patch-1
Jan 24, 2026
Merged

Allow specifying datasets in configs, change default optimizer, improved CUDA support#1
aronson merged 6 commits intomainfrom
fizzAI-patch-1

Conversation

@fizzAI
Copy link
Member

@fizzAI fizzAI commented Jan 23, 2026

AdamW is one morbillion times better than Adam

@fizzAI fizzAI changed the title Switch to AdamW, add boilerplate for more optimizers later Switch to AdamW, add boilerplate for more optimizers later, update Python dep to 3.12 Jan 23, 2026
@fizzAI fizzAI marked this pull request as draft January 23, 2026 23:00
@fizzAI fizzAI changed the title Switch to AdamW, add boilerplate for more optimizers later, update Python dep to 3.12 Allow specifying datasets in configs, change default optimizer, improved CUDA support Jan 23, 2026
@fizzAI fizzAI marked this pull request as ready for review January 23, 2026 23:39
@aronson aronson self-requested a review January 23, 2026 23:42
Copy link
Member

@aronson aronson left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

looks really good, I need to take ~30 minutes to run 2-3 tests on my mac to make sure we still have feature parity. if you could, @fizzAI can you post a screenshot of the application working on your CUDA setup? also name tersely which CUDA version and card you're using for posterity. then I can merge np

@aronson aronson self-assigned this Jan 23, 2026
@aronson
Copy link
Member

aronson commented Jan 23, 2026

LGTM over here

Screen.Recording.2026-01-23.at.5.51.05.PM.mov

@fizzAI
Copy link
Member Author

fizzAI commented Jan 23, 2026

image

Nvidia GTX 1650 Super, Cuda 12 MLX and Cuda runtime

@fizzAI fizzAI requested a review from aronson January 23, 2026 23:59
Signed-off-by: Heni <i@pingas.org>
@aronson aronson merged commit b1e2523 into main Jan 24, 2026
1 check passed
@aronson aronson deleted the fizzAI-patch-1 branch January 24, 2026 00:05
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants