-
Notifications
You must be signed in to change notification settings - Fork 91
Pull requests: quic/efficient-transformers
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Reduce MoE export RAM by aliasing fused expert weights as views
#1121
opened Jun 24, 2026 by
quic-rishinr
Contributor
Loading…
cleaned up layerwise API and added CustomLoader (#1105)
#1120
opened Jun 24, 2026 by
ochougul
Contributor
Loading…
Make num_cores (NSP) configurable via NUM_CORES env for nightly pipeline
#1117
opened Jun 24, 2026 by
quic-vishali
Contributor
Loading…
fix(QRANIUMSW-62219): Gemma4 pop vision_size from compiler_options in…
#1116
opened Jun 24, 2026 by
quic-gthiruko
•
Draft
RepeatKV Transform changes with comments addressed
#1114
opened Jun 24, 2026 by
quic-dhirajku
Contributor
Loading…
Updated supported models list in validate.md
#1112
opened Jun 23, 2026 by
quic-vishali
Contributor
Loading…
fix(mixtral): fix MXFP6 quantization and ONNX export issues for Mixtral_moe
#1107
opened Jun 22, 2026 by
Cs23m011
Loading…
Added fix to layerwise implementation to incorporate CB
#1106
opened Jun 22, 2026 by
abhishek-singh591
Contributor
Loading…
Subfunction fix w/o Layerwise
1.22
Release 1.22 candidate
bugfix
#1104
opened Jun 19, 2026 by
mohiso22
Contributor
Loading…
Fix(0618)(layerwise): dedup merged ONNX graph and keep key prefill passess enabled.
1.22
Release 1.22 candidate
bugfix
#1099
opened Jun 18, 2026 by
vbaddi
Contributor
Loading…
docs: Add automated MkDocs documentation system with versioning
#1095
opened Jun 18, 2026 by
quic-amitraj
Contributor
•
Draft
KV handoff with buffer slicing APIs to avoid KV I/O copies
enhancement
New feature or request
#1087
opened Jun 16, 2026 by
quic-akuruvil
Contributor
Loading…
Previous Next
ProTip!
Find all pull requests that aren't related to any open issues with -linked:issue.