Skip to content

Pull requests: AI-Hypercomputer/maxtext

New pull request New
177 Open 3,081 Closed
177 Open 3,081 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

Support option to skip the optimizer for training step gemini-review
#3490 opened Mar 24, 2026 by RissyRan Loading…
4 tasks done
@gagika
9
Add checkpoint deletion options to configuration and Orbax checkpoint manager
#3488 opened Mar 23, 2026 by awonak Loading…
3 of 4 tasks
Fix: Upload docker images on pypi release
#3487 opened Mar 23, 2026 by SurbhiJainUSC Loading…
4 tasks done
1
[Test] Rope change
#3486 opened Mar 23, 2026 by RissyRan • Draft
4 tasks
1
Add option to replicate attn weights on FSDP or EP
#3480 opened Mar 21, 2026 by gobbleturk Loading…
1
[Distillation] Fix eod masking + strategy refactoring pull ready
#3478 opened Mar 20, 2026 by vlad-karp Loading…
4 tasks done
2
Support modern uv project management (uv init + uv add)
#3474 opened Mar 20, 2026 by igorts-git • Draft
4 tasks done
1
1
Split logical names in moe module pull ready
#3473 opened Mar 20, 2026 by NuojCheng Loading…
4 tasks done
10
Fix DiLoCo training compatibility issues
#3471 opened Mar 20, 2026 by khatwanimohit Loading…
4 tasks done
5
NNX migration prep (2/N): NNX utils and sharding utilities
#3470 opened Mar 20, 2026 by ecnal-cienet • Draft
4 tasks done
1
Fix macOS installation by dynamically skipping legacy MaxText directory
#3469 opened Mar 20, 2026 by igorts-git Loading…
4 tasks done
1
3
Remove legacy uppercase MaxText from src, to ensure maxtext works on MacOS which has a case-insensitive filesystem.
#3467 opened Mar 20, 2026 by syrkis Loading…
4 tasks done
1
Fix param mapping for Qwen2 gemini-review
#3466 opened Mar 20, 2026 by ChingTsai Loading…
4 tasks done
7
Fix src/MaxText references in GPU/runner Dockerfiles pull ready
#3462 opened Mar 20, 2026 by bvandermoon Loading…
4 tasks done
1
Integrating b/v blocking for linear softmax cross entropy loss in maxtext
#3461 opened Mar 19, 2026 by copybara-service bot Loading…
1
maxtext/tunix lora integration [WIP]
#3453 opened Mar 18, 2026 by andytwigg • Draft
4 tasks
1
Add simplified APIs for model obtaining maxtext models
#3450 opened Mar 18, 2026 by A9isha Loading…
4 tasks done
1
modify process_data to generate separate user/system parts in prompts
#3445 opened Mar 18, 2026 by andytwigg Loading…
4 tasks done
4
let tunix see lr hyperparams for logging pull ready
#3444 opened Mar 18, 2026 by andytwigg Loading…
4 tasks done
3
add exact match reward, make_reward_fn wrapper, remove check_answer from default rewards
#3443 opened Mar 18, 2026 by andytwigg Loading…
4 tasks done
7
NNX train
#3442 opened Mar 18, 2026 by charlesli640 • Draft
4 tasks done
3
Add abort_on_nan_loss and abort_on_inf_loss options
#3440 opened Mar 18, 2026 by Steboss Loading…
4 tasks done
5
Added SFT Pre-Processing for Grain Input Pipeline
#3437 opened Mar 18, 2026 by ajkv-google Loading…
4 tasks done
18
Use non-tokamax group sizes for expert parallelism
#3436 opened Mar 18, 2026 by xuefgu • Draft
4 tasks done
1
Use Tokamax's representative group sizes.
#3434 opened Mar 17, 2026 by copybara-service bot Loading…
1
ProTip! What’s not been updated in a month: updated:<2026-02-24.