-
Notifications
You must be signed in to change notification settings - Fork 448
Pull requests: InternLM/xtuner
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
[Fix] Continue VLM M-RoPE response positions from global amax
#2097
opened Sep 18, 2026 by
YanhuiDua
Collaborator
Loading…
1 of 2 tasks
[Feature] Decouple expert parallelism from FSDP sharding (dp2ep layout)
#2093
opened Sep 17, 2026 by
silencelamb
Loading…
Fix/empty cache and refine transport
#2092
opened Sep 17, 2026 by
PengchengShi00
Collaborator
Loading…
[Feat] Fuse MoE activation FP8 quantization kernels
#2089
opened Sep 15, 2026 by
qiyueliuhuo23
Loading…
[ CI ] Add recomputation cases for GLM‑5.2
#2086
opened Sep 15, 2026 by
kkscilife
Collaborator
Loading…
[WIP] Refactor prepare_train_data to reduce centralized data transfer
#2084
opened Sep 11, 2026 by
YanhuiDua
Collaborator
Loading…
3 tasks
[Feature] Add optional FlashMLA backend for DSA Sparse MLA
#2083
opened Sep 11, 2026 by
qiyueliuhuo23
Loading…
feat: add FlashMLA forward-only backend for sparse MLA
#2081
opened Sep 9, 2026 by
PengchengShi00
Collaborator
Loading…
feat(tools): add HF model normalization utility
#2073
opened Sep 6, 2026 by
caixianzhang
Collaborator
Loading…
[Fix] Update flush cache worker broadcast
#2071
opened Sep 5, 2026 by
matrix72c
Contributor
Loading…
[Feature] Add end-to-end TV loss for MTP SFT
#2070
opened Sep 4, 2026 by
JT-Ushio
Contributor
Loading…
[Feature] Add query-chunked TileLang DSA selector
#2068
opened Sep 4, 2026 by
Denny991
Collaborator
Loading…
[Fix] Transfer routed experts release ownership
#2064
opened Sep 3, 2026 by
matrix72c
Contributor
Loading…
[Fix] alpaca_enzh_dataset passes a shuffle_before_pack kwarg neither callee accepts
#2057
opened Aug 29, 2026 by
Anai-Guo
Loading…
3 tasks done
[Feature] Integrate MoonEP dispatcher for FSDP expert-parallel training
#2056
opened Aug 28, 2026 by
jayhenry
Collaborator
Loading…
[WIP] Add index-based data packing and worker-side materialization
#2055
opened Aug 28, 2026 by
YanhuiDua
Collaborator
Loading…
[Feature] Support Intern-S2-Preview training on NPU
#2052
opened Aug 27, 2026 by
CyCle1024
Collaborator
Loading…
[Feature] Integrate UltraEP Runtime for MoE EP Load Balancing
#2050
opened Aug 26, 2026 by
ShilohYu
Collaborator
Loading…
Previous Next
ProTip!
no:milestone will show everything without a milestone.