-
Notifications
You must be signed in to change notification settings - Fork 35
Pull requests: intel/auto-round
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Initial assessment of alternative double quantization tuning strategies
#535
opened Apr 22, 2025 by
SinpackKonmakan
•
Draft
enable llama4 int8 quantization baseline
draft
wontfix
This will not be worked on
#522
opened Apr 16, 2025 by
WeiweiZhang1
Loading…
[not4land] temp change to convert torchao checkpoint to gguf
#504
opened Apr 11, 2025 by
jerryzh168
Loading…
ProTip!
Exclude everything labeled
bug
with -label:bug.