Skip to content

Pull requests: AI-Hypercomputer/maxtext

New pull request New
161 Open 3,476 Closed
161 Open 3,476 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

Feature: add standalone DPO training and data hooks
#3883 opened May 12, 2026 by igorts-git Collaborator Loading…
4 tasks done
1
Plumbing and core MoE logic for router replay
#3881 opened May 12, 2026 by xuefgu Collaborator • Draft
4 tasks done
4
Use env credentials instead of passing them through
#3880 opened May 12, 2026 by bvandermoon Collaborator Loading…
4 tasks done
1
Improve error message when tokenize_data config doesn't match dataset
#3879 opened May 12, 2026 by aireenmei Collaborator Loading…
4 tasks done
1
Publish TPU pre-training and post-training stable docker images to cloud-tpu-images
#3878 opened May 11, 2026 by SurbhiJainUSC Collaborator • Draft
4 tasks done
1
Secure deserialization and model checkpoint loading pipelines
#3877 opened May 11, 2026 by bvandermoon Collaborator Loading…
4 tasks done
1
enable aot identification test
#3876 opened May 11, 2026 by NuojCheng Collaborator • Draft
4 tasks
1
Improve performance of gemma4 MoE inference.
#3875 opened May 11, 2026 by NicoGrande Collaborator Loading…
4 tasks done
7
WIP DO NOT REVIEW Enable sft_trainer_correctness_test
#3874 opened May 11, 2026 by igorts-git Collaborator • Draft
4 tasks
1
write dequantization scripts for DeepSeek V4 FP4/FP8 weights
#3873 opened May 11, 2026 by snehalv2002 Collaborator Loading…
4 tasks done
1
Add zero1 aot support in train compile
#3872 opened May 11, 2026 by NuojCheng Collaborator • Draft
4 tasks done
1
Implement custom MoE HashRouter, TopKRouter, and sqrtsoftplus
#3871 opened May 11, 2026 by parambole Collaborator • Draft
4 tasks
1
Conditionally branch tokamax.ragged_dot calls based on use_manual_quantization
#3869 opened May 11, 2026 by zxhe-sean Collaborator Loading…
4 tasks done
1
DeepSeek V4 Integration
#3867 opened May 11, 2026 by parambole Collaborator • Draft
4 tasks
1
Implement DeepSeek-V4 Compressed Attention Layers
#3866 opened May 11, 2026 by parambole Collaborator • Draft
4 tasks
1
DeepSeek-V4 Core Primitives
#3865 opened May 11, 2026 by parambole Collaborator • Draft
4 tasks
1
[DeepSeek v3] Add grad mask and update MLA init gemini-review
#3864 opened May 10, 2026 by gagika Collaborator Loading…
4 tasks done
9
Enable Qwen3-Omni SFT on ChartQA
#3863 opened May 10, 2026 by hengtaoguo Collaborator • Draft
4 tasks
1
Optimize MaxText unit and integration test suite runtime
#3860 opened May 9, 2026 by shralex Collaborator Loading…
4 tasks done
4
Update optimization docs and add TPU v7x guide
#3857 opened May 8, 2026 by jacoguzo Collaborator Loading…
4 tasks done
11
Update docker image guide
#3855 opened May 8, 2026 by melissawm Collaborator Loading…
1 task done
Update JAX to 0.10.0 for pre-training
#3854 opened May 8, 2026 by SurbhiJainUSC Collaborator • Draft
4 tasks done
1
Update First Run tutorial
#3853 opened May 8, 2026 by melissawm Collaborator Loading…
1 task done
Trigger tests using PR comments
#3850 opened May 8, 2026 by shralex Collaborator Loading…
4 tasks done
2
[NNX] NNX migration prep (10/N): vocab tiling custom_vjp with output-head carve-out
#3849 opened May 8, 2026 by ecnal-cienet Collaborator • Draft
4 tasks done
1
ProTip! Type g i on any issue or pull request to go back to the issue listing page.