Skip to content

Pull requests: ISEEKYAN/Megatron-LM

New pull request New
11 Open 58 Closed
11 Open 58 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
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Fix MLite microbatch loss and forward-only output contracts
#68 opened Jun 28, 2026 by ISEEKYAN Owner Loading…
[lite] Harden GLM5.2 readiness and checkpoint recovery (#66 follow-up)
#67 opened Jun 27, 2026 by Meirtz • Draft
[lite] GLM5.2 (DeepSeek-V3.2) IndexShare DSA support
#66 opened Jun 27, 2026 by ISEEKYAN Owner Loading…
[codex] DSv4: derive config from HF + legacy-load + exporter DTensor materialize
#65 opened Jun 23, 2026 by Meirtz • Draft
1
[codex] DSv4 DSA: opt-in torch fallback paths for smoke/CI (default-off)
#64 opened Jun 23, 2026 by Meirtz Loading…
1
[codex] Fix DSv4 DCP checkpoint placements for DTensor-like params
#61 opened Jun 21, 2026 by Meirtz Loading…
ds4 (deepseek_v4): DSA indexer aux-loss scale hook + lift forced CP=1 (MTP/dense-CSA support CP>1)
#59 opened Jun 18, 2026 by ISEEKYAN Owner Loading…
mlite: unified MoE router replay across 5 models
#49 opened Jun 16, 2026 by ISEEKYAN Owner Loading…
Add Megatron-FSDP optimizer primitive to MLite
#26 opened Jun 11, 2026 by ISEEKYAN Owner Loading…
Add chunked-EP MoE primitive
#13 opened Jun 8, 2026 by ISEEKYAN Owner Loading…
Add Megatron-FSDP (mfsdp) optimizer backend
#12 opened Jun 8, 2026 by ISEEKYAN Owner Loading…
ProTip! Type g p on any issue or pull request to go back to the pull request listing page.