Skip to content

Pull requests: AI-Hypercomputer/maxtext

New pull request New
164 Open 3,304 Closed
164 Open 3,304 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

Save run manifest for distillation reproducibility. gemini-review
#3709 opened Apr 21, 2026 by gagika Collaborator Loading…
4 tasks done
5
Add optional --skip-validation flag to benchmark recipes and XPK workload creation
#3708 opened Apr 21, 2026 by RUEI4341 Contributor Loading…
4 tasks done
[Draft] Onboard ragged gather to moe draft Draft PR
#3707 opened Apr 21, 2026 by NuojCheng Collaborator • Draft
4 tasks
1
[sc] Removed usages of "sc.persistent" in SparseCore kernels
#3704 opened Apr 20, 2026 by copybara-service bot Loading…
1
draft pull ready
#3703 opened Apr 20, 2026 by shuningjin Collaborator • Draft
4 tasks done
1
Jackyf/qlora
#3702 opened Apr 20, 2026 by RexBearIU Collaborator • Draft
4 tasks
distillation: resume + xpk launcher + metrics refactor gemini-review
#3701 opened Apr 20, 2026 by gagika Collaborator Loading…
4 tasks done
12
Speed up cpu-unit CI gemini-review
#3700 opened Apr 19, 2026 by gagika Collaborator • Draft
4 tasks
7
fix masking error when using mlp_bias=True causing NaN during gradien…
#3699 opened Apr 19, 2026 by snehalv2002 Collaborator Loading…
4 tasks done
3
Offload all checkpoints to host and remove unused mesh axes.
#3697 opened Apr 17, 2026 by copybara-service bot Loading…
1
Adds ability to pass chat_template_path argument for SFT to work with non-instruct models gemini-review
#3696 opened Apr 17, 2026 by SurbhiJainUSC Collaborator Loading…
4 tasks done
15
update test
#3690 opened Apr 17, 2026 by shuningjin Collaborator • Draft
4 tasks
1
Enable elastic training for RL
#3689 opened Apr 17, 2026 by copybara-service bot Loading…
@abhinavclemson
1
[Distillation] base learn-to-init llama attention for distillation
#3688 opened Apr 17, 2026 by vlad-karp Collaborator Loading…
4 tasks done
1
[DRAFT] Vllm fused moe draft Draft PR
#3687 opened Apr 16, 2026 by NuojCheng Collaborator • Draft
4 tasks
feat: Add FLOPs calculation for Multi-Token Prediction (MTP) modules gemini-review
#3685 opened Apr 16, 2026 by parambole Collaborator Loading…
4 tasks done
8
[DO NOT REVIEW] Update generated requirements for post-training with JAX 0.9.2
#3684 opened Apr 16, 2026 by SurbhiJainUSC Collaborator • Draft
4 tasks done
1
docs: add LoRA tutorial and layer customization guide
#3682 opened Apr 16, 2026 by chiajunglien • Draft
4 tasks
11
[Inference] Diverse Beam Search Integration
#3681 opened Apr 16, 2026 by yipkingster Loading…
5 tasks done
1
Add MoE load balancing loss to distillation
#3679 opened Apr 16, 2026 by JamesDeng42 Collaborator Loading…
4 tasks done
7
fix gemma4 to vllm weight conversion
#3677 opened Apr 15, 2026 by aireenmei Collaborator • Draft
4 tasks
1
test: refactor DeepSeek v3 MTP tests into standard two-step structure pull ready
#3676 opened Apr 15, 2026 by parambole Collaborator Loading…
4 tasks done
12
Adds ability to pass chat_template_path argument into MaxText SFT, loading a separate chat_template from the tokenizer that's provided.
#3675 opened Apr 15, 2026 by copybara-service bot Loading…
4 tasks done
1
Support grain data checkpoint for elastic training
#3673 opened Apr 15, 2026 by aireenmei Collaborator • Draft
4 tasks done
1
Make all links internal (where possible)
#3671 opened Apr 15, 2026 by melissawm Collaborator Loading…
1 task done
ProTip! Follow long discussions with comments:>50.