Skip to content

Pull requests: pytorch/torchtitan

New pull request New
Clear current search query, filters, and sorts
221 Open 1,626 Closed
221 Open 1,626 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

[WIP] M*G Triton group gemm for MoE training CLA Signed This label is managed by the Meta Open Source bot.
#967 by lessw2020 was merged Apr 3, 2025 Loading…
11 🚀 5
[WIP] Integrate DeepGEMM, add supporting utils and unit testing, to enable blockwise fp8 inference CLA Signed This label is managed by the Meta Open Source bot.
#1124 by lessw2020 was merged Apr 24, 2025 Loading…
59 🚀 5
[WIP] Experimental implementation of gpt-oss (grouped GEMM MoE + FlexAttention sink/sliding) CLA Signed This label is managed by the Meta Open Source bot.
#1559 by KhoomeiK was closed Oct 23, 2025 Loading…
37 🚀 4
Add deterministic RL training experiment with vLLM CLA Signed This label is managed by the Meta Open Source bot.
#1975 by bwasti was merged Nov 7, 2025 Loading…
35 🚀 3
[DeepSeek] Let MoEs share Symmetric Memory across layers CLA Signed This label is managed by the Meta Open Source bot.
#958 by kwen2501 was closed Mar 20, 2025 Loading…
2 🚀 3
gpt-oss model enablement CLA Signed This label is managed by the Meta Open Source bot.
#1754 by wwwjn was merged Oct 22, 2025 Loading…
66 🚀 3
add benchmarks folder and submission guidelines CLA Signed This label is managed by the Meta Open Source bot.
#1296 by tianyu-l was merged Jun 13, 2025 Loading…
1 🚀 3
Add the option to turn on async-TP CLA Signed This label is managed by the Meta Open Source bot.
#429 by yifuwang was merged Jun 27, 2024 Loading…
12 🚀 2
[Qwen3] Qwen3 MoE initial support CLA Signed This label is managed by the Meta Open Source bot.
#1685 by wwwjn was merged Sep 11, 2025 Loading…
30 🚀 2
[405B] Add performance data for 405B model CLA Signed This label is managed by the Meta Open Source bot.
#554 by fduwjj was merged Aug 23, 2024 Loading…
20 🚀 1
[HF] Model Definition Conversion Support for FLUX CLA Signed This label is managed by the Meta Open Source bot.
#1582 by wesleytruong was merged Aug 20, 2025 Loading…
8 🚀 1
[Kernels] add triton contiguous groupgemm CLA Signed This label is managed by the Meta Open Source bot.
#1154 by lessw2020 was merged May 1, 2025 Loading…
12 🚀 1
CUDAGraph support for SimpleFSDP and TP CLA Signed This label is managed by the Meta Open Source bot.
#2050 by BoyuanFeng was merged Nov 20, 2025 Loading…
4 tasks done
22 🚀 1
add llama4 as an experiment CLA Signed This label is managed by the Meta Open Source bot.
#1064 by tianyu-l was merged Apr 7, 2025 Loading…
🚀 1
upgrade generate_permute_indices with faster kernel (parallelized) CLA Signed This label is managed by the Meta Open Source bot.
#1098 by lessw2020 was merged Apr 15, 2025 Loading…
4 🚀 1
[WIP] Used per-parameter FSDP CLA Signed This label is managed by the Meta Open Source bot.
#70 by awgu was closed Mar 13, 2024 • Draft
2 of 5 tasks
6 🚀 1
[llama4] fall back to for-loop based MoE if not on SM90 or later CLA Signed This label is managed by the Meta Open Source bot.
#1096 by tianyu-l was merged Apr 12, 2025 Loading…
1
2 🚀 1
[DeepSeek] Integrate M*G Group Gemm CLA Signed This label is managed by the Meta Open Source bot.
#1046 by kwen2501 was merged Apr 3, 2025 Loading…
1 🚀 1
[DeepSeek] all-to-all-v kernel writes out output_splits CLA Signed This label is managed by the Meta Open Source bot.
#941 by kwen2501 was merged Mar 20, 2025 Loading…
23 🚀 1
Implement async_checkpoint CLA Signed This label is managed by the Meta Open Source bot.
#313 by fegin was merged May 7, 2024 Loading…
3 🚀 1
enable gc control scheduling to help avoid stragglers CLA Signed This label is managed by the Meta Open Source bot.
#148 by lessw2020 was merged Mar 20, 2024 Loading…
27 🚀 1
[MoE] DeepEP refactor and fix memory leak during training and inference ciflow/8gpu CLA Signed This label is managed by the Meta Open Source bot.
#2296 by shuhuayu was merged Jan 29, 2026 Loading…
10 🚀 1
Use cudnn backend on B200 CLA Signed This label is managed by the Meta Open Source bot.
#1196 by drisspg was merged May 15, 2025 Loading…
3 🚀 1
add L40s gpu type for MFU calcs CLA Signed This label is managed by the Meta Open Source bot.
#1204 by lessw2020 was merged May 19, 2025 Loading…
🚀 1
[BC Breaking] Config System Refactor: TOML to Python Dataclass Registry ciflow/8gpu CLA Signed This label is managed by the Meta Open Source bot.
#2386 by tianyu-l was merged Feb 23, 2026 Loading… New Feature, Model, Misc
@tianyu-l
126 🚀 1
ProTip! Exclude everything labeled bug with -label:bug.