-
Notifications
You must be signed in to change notification settings - Fork 11
Pull requests: JoshEngels/MultiDimensionalFeatures
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Remove NVIDIA CUDA 12.1 hard pins from requirements
#24
opened Mar 13, 2026 by
Tauheed-Elahee
Loading…
Fix incorrect --clustering_type arg in README examples
#20
opened Mar 13, 2026 by
Tauheed-Elahee
Loading…
Add compatibility for newer sae_lens forward() return type
#19
opened Mar 13, 2026 by
Tauheed-Elahee
Loading…
Use dynamic device count instead of hardcoded num_devices=2
#18
opened Mar 13, 2026 by
Tauheed-Elahee
Loading…
Update torch.load calls to include weights_only parameter
#16
opened Mar 12, 2026 by
Tauheed-Elahee
Loading…
Revert to cpu if cuda is not available (simpler version)
#11
opened Sep 9, 2024 by
JasonGross
Contributor
Loading…
Add support for loading 16-bit models with
--dtype
#10
opened Sep 9, 2024 by
JasonGross
Contributor
Loading…
Revert to cpu if cuda is not available (backwards compatible version)
#7
opened Sep 7, 2024 by
JasonGross
Contributor
Loading…
Fix to better support MPS: convert to float32 earlier
#6
opened Sep 7, 2024 by
JasonGross
Contributor
Loading…
Use pathlib more and have default cache paths be relative to the repo directory
#5
opened Sep 7, 2024 by
JasonGross
Contributor
Loading…
Don't fail with "AssertionError: Not enough CUDA devices to support n_devices 2" when there aren't 2 GPUs
#4
opened Aug 28, 2024 by
JasonGross
Contributor
Loading…
ProTip!
no:milestone will show everything without a milestone.