Skip to content

Issues: linkedin/Liger-Kernel

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

52 Open 89 Closed
52 Open 89 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
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

Dtype Mismatch in torch.addmm within ops/fused_linear_cross_entropy.py in AMP training
#501 opened Dec 26, 2024 by DandinPower
Extending Liger-Kernel Optimizations to Encoder Models Like BER
#500 opened Dec 26, 2024 by pengzhangzhi
error when run sh run_qwen.sh good first issue Good for newcomers
#487 opened Dec 18, 2024 by CharlesJhonson
3
What's the different with FlagGems
#478 opened Dec 14, 2024 by CharlesJhonson
Potential Optimization for Preference Training with Prefix Sharing
#476 opened Dec 13, 2024 by austin362667
Support offline logits for teacher model
#441 opened Dec 9, 2024 by austin362667
@austin362667
Bug in chunking?
#439 opened Dec 8, 2024 by cinjon
10
PreferenceBase with Softcapping?
#426 opened Dec 4, 2024 by cinjon
1
2
The accuracy is misaligned when using bf16
#414 opened Nov 28, 2024 by starstream
No Significant Improvement Observed in Model Training Speed
#409 opened Nov 27, 2024 by lianghsun
4
Weighted Cross Entropy Loss
#404 opened Nov 21, 2024 by winglian
2
3
ValueError: Pointer argument (at 0) cannot be accessed from Triton (cpu tensor?)
#401 opened Nov 20, 2024 by shivam15s
1
Apple's cross entropy computation
#391 opened Nov 17, 2024 by fzyzcjy
8
[RFC] Liger FlexChunkLoss: Alignment and Distillation loss
#371 opened Nov 8, 2024 by shivam15s
5 of 12 tasks
@shivam15s
21
cross entropy kernel: inplace modification of input
#343 opened Nov 4, 2024 by mgrabban
2
Overlap matrix multiplication (needs tensor core) and other things like activation (needs cuda core and memory bandwidth) to speed up
#341 opened Nov 4, 2024 by fzyzcjy
Possible support for weighted average loss calculation in FusedLinearCrossEntropy kernel
#338 opened Nov 1, 2024 by ChenlongDeng
FSDP fails with [ lm_head ] layer
#332 opened Oct 31, 2024 by gotzmann
5
Training LLaVA with the Liger kernel results in degraded performance.
#319 opened Oct 22, 2024 by y-rok
3
mllama patch modifies nn.LayerNorm globally
#315 opened Oct 19, 2024 by tyler-romero
3
Empty Medusa head tensors
#309 opened Oct 14, 2024 by vkc1vk
2
Add end-to-end example of mllama
#301 opened Oct 10, 2024 by shivam15s
@shivam15s
1
2024 Q4 Roadmap
#285 opened Sep 30, 2024 by ByronHsu
Add TVD (Total variation distance) Kernel feature
#281 opened Sep 28, 2024 by qingquansong
1
@saurabhkoshatwar
3
In-place operations in triton kernel might result in incorrect gradient calculations bug Something isn't working
#272 opened Sep 26, 2024 by Tcc0403
1
3
ProTip! no:milestone will show everything without a milestone.