Skip to content

Pull requests: AI-Hypercomputer/maxtext

New pull request New
235 Open 4,122 Closed
235 Open 4,122 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

Run selected tests in Decoupled mode on CPU in CI
#4642 opened Jul 28, 2026 by shralex Collaborator Loading…
4 tasks done
1
Add E2E TPU testing scripts for GPTOSS-20B
#4641 opened Jul 28, 2026 by chiajunglien Contributor Loading…
4 tasks done
2
first change
#4640 opened Jul 28, 2026 by mesakhcienet Collaborator • Draft
4 tasks
1
Migrate deprecated XLA collective pipelining flags in MaxText
#4639 opened Jul 28, 2026 by copybara-service Bot Loading…
1
MHC Projection Fusion
#4638 opened Jul 28, 2026 by muskansh-google • Draft
4 tasks
1
[DECOUPLED-MODE] checkpoint_conversion: import GCS via gcloud_stub in load_dynamic
#4637 opened Jul 28, 2026 by gulsumgudukbay Collaborator Loading…
6 tasks done
2
[MaxText][RL] Omit Redundant Weight Resharding Prior to Step 0 Evaluation & Add Engram Inheritance
#4636 opened Jul 27, 2026 by igorts-git Collaborator • Draft
4 tasks done
1
experiment
#4635 opened Jul 27, 2026 by Shuwen-Fang Collaborator Loading…
4 tasks
1
Fix DeepSeek decode running the prefill attention path under NNX
#4634 opened Jul 27, 2026 by ecnal-cienet Collaborator • Draft
4 tasks done
3
Refactor layer-wise tests with torch
#4633 opened Jul 27, 2026 by hengtaoguo Collaborator Loading…
4 tasks done
3
Add ep to moe test for chunking
#4631 opened Jul 27, 2026 by Shuwen-Fang Collaborator Loading…
4 tasks done
2
Add flag-controlled sliced MLA projections
#4630 opened Jul 27, 2026 by dandragona Collaborator Loading…
4 tasks done
2
Introduce an inflight computation throttler to rate-limit TPU computations in MaxText training engine.
#4629 opened Jul 27, 2026 by copybara-service Bot Loading…
@SurbhiJainUSC
1
Add rope_pairwise config flag to optimize YaRN RoPE
#4627 opened Jul 27, 2026 by dandragona Collaborator Loading…
3 tasks done
2
Optimize TPU Tokamax ring backward
#4626 opened Jul 27, 2026 by huytransformer Collaborator • Draft
4 tasks
2
feat(input): Megatron MMap Data Formats for MaxText Grain
#4625 opened Jul 27, 2026 by Yitrus Loading…
4 tasks done
1
fix: MTP CP-aware roll, segment-aware roll, and synthetic packing support
#4623 opened Jul 27, 2026 by chiaotung97 Loading…
6 tasks done
2
Support packed TPU Tokamax ring attention
#4622 opened Jul 27, 2026 by huytransformer Collaborator Loading…
4 tasks done
3
fix(nnx): Fix Linen checkpoint restoration into NNX Gemma4 model
#4619 opened Jul 27, 2026 by hsuan-lun-chiang Collaborator • Draft
1
Support vllm_decode for Qwen3-VL
#4618 opened Jul 27, 2026 by hengtaoguo Collaborator • Draft
4 tasks
2
fix: fallback to dot_product for autoselected attention on CPU to prevent Tokamax NotImplementedError pull ready
#4617 opened Jul 27, 2026 by chiajunglien Contributor Loading…
4 tasks done
6
Fix eval and checkpointing under load-balanced CP
#4613 opened Jul 26, 2026 by huytransformer Collaborator • Draft
4 tasks
2
Support MLA TPU Tokamax ring attention
#4610 opened Jul 25, 2026 by huytransformer Collaborator • Draft
4 tasks
2
End2End Video SFT with Varied Size/Duration
#4608 opened Jul 24, 2026 by hengtaoguo Collaborator • Draft
4 tasks
2
Fix logical to physical axis mapping for attention heads.
#4607 opened Jul 24, 2026 by copybara-service Bot Loading…
4 tasks done
@niting
1
ProTip! Exclude everything labeled bug with -label:bug.