Skip to content

Pull requests: linkedin/Liger-Kernel

New pull request New
30 Open 463 Closed
30 Open 463 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Add support for Qwen3 MoE models
#706 opened May 8, 2025 by chiwanpark Loading…
3 tasks done
1
3
Allow option to choose between fused LCE and CE loss
#704 opened May 6, 2025 by connermanuel Loading…
2 of 3 tasks
1
Solar pro implementation
#699 opened May 4, 2025 by KareemMusleh Loading…
2 of 4 tasks
3
Fix empty Medusa head tensors
#698 opened May 3, 2025 by chiwanpark Loading…
1 of 3 tasks
1
Multi-token Attention
#689 opened Apr 28, 2025 by AndreSlavescu Loading…
3 tasks done
Sparsemax
#687 opened Apr 28, 2025 by AndreSlavescu Loading…
3 tasks done
8
Make FLCE compatible with FSDP and PEFT
#674 opened Apr 21, 2025 by astefanutti Loading…
1 of 3 tasks
2
5
element-wise-DyT faster than the origin LigerDyT
#673 opened Apr 21, 2025 by mdy666 Loading…
11
GRPO Loss kernel fully write by triton, reduce 46G memory
#672 opened Apr 19, 2025 by mdy666 Loading…
8
Modified block and warp sizes for improved performance on XPU for both layernnorm and rmsnorm
#661 opened Apr 10, 2025 by Tarakarevu1 Loading…
3 tasks done
4
[WIP] Update benchmark data
#643 opened Apr 2, 2025 by Tcc0403 • Draft
18 of 23 tasks
6
Fix incorrect module patching when using LoRA with modules_to_save
#632 opened Mar 26, 2025 by BenasdTW Loading…
3 tasks done
1
Fix: pailgemma type error
#627 opened Mar 25, 2025 by jp1924 Loading…
3 tasks
8
[DRAFT] Add deepseek v3 monkey patch
#583 opened Feb 24, 2025 by zcnrex • Draft
3 tasks
2
Fix Setup.py about "building from source on ROCM"
#572 opened Feb 20, 2025 by hebiao064 Loading…
3 tasks
1
6
add batch_norm op with test and benchmark
#559 opened Feb 7, 2025 by yanghailong-git Loading…
3 tasks done
2
Add Flex Attention Monkey Patch for LLAMA
#540 opened Jan 25, 2025 by austin362667 Loading…
2 of 3 tasks
[Model] DeepseekV2 Support
#499 opened Dec 26, 2024 by saurabhkoshatwar Loading…
3 tasks done
1
16
Refactor chunked preference functions and distillation base class
#491 opened Dec 20, 2024 by shivam15s Loading…
1 of 3 tasks
2
test: Add test for ref_input parameter in fused linear preference
#468 opened Dec 11, 2024 by xingyaoww • Draft
Add on-paper form of RoPE kernel
#465 opened Dec 10, 2024 by Comet0322 Loading…
2 of 3 tasks
1
Implement softcapping in fused jsd
#403 opened Nov 21, 2024 by wheynelau Loading…
2 of 3 tasks
3
[WIP]: Autotune Chunk Size
#395 opened Nov 19, 2024 by pramodith • Draft
2 of 3 tasks
poke ci
#350 opened Nov 5, 2024 by ByronHsu Loading…
3 tasks
added batch norm
#321 opened Oct 22, 2024 by vulkomilev Loading…
ProTip! no:milestone will show everything without a milestone.