Skip to content

Pull requests: pytorch/torchtitan

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
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort

Pull requests list

[Experimental Feature] Huggingface model training CLA Signed This label is managed by the Meta Open Source bot.
#919 opened Mar 3, 2025 by junjzhang Loading…
IBM experimental dataloaders CLA Signed This label is managed by the Meta Open Source bot.
#376 opened May 31, 2024 by daviswer Loading…
Execute moe.gate in float32 CLA Signed This label is managed by the Meta Open Source bot. high priority
#2389 opened Feb 17, 2026 by chelsea0x3b Loading… New Feature, Model, Misc
[torchtitan][debug] integrated CommDebugMode into TorchTitan CLA Signed This label is managed by the Meta Open Source bot.
#480 opened Jul 24, 2024 by sinhaanshul Loading…
[DO NOT REVIEW] debug fsdp2 checkpoint for uneven sharding CLA Signed This label is managed by the Meta Open Source bot.
#1635 opened Aug 25, 2025 by weifengpy Draft
Remove device to host synchronizations from repeat_interleave and tail_slack CLA Signed This label is managed by the Meta Open Source bot.
#2440 opened Feb 25, 2026 by rthekini-aws Loading…
Distributed Scion/Muon CLA Signed This label is managed by the Meta Open Source bot.
#1630 opened Aug 25, 2025 by rakkit Loading…
Staging SFT training CLA Signed This label is managed by the Meta Open Source bot.
#2148 opened Dec 12, 2025 by rakkit Loading…
[WIP] zero bubble CLA Signed This label is managed by the Meta Open Source bot.
#546 opened Aug 20, 2024 by H-Huang Draft
Add shims for PyTorch stable CLA Signed This label is managed by the Meta Open Source bot.
#1862 opened Oct 13, 2025 by bwasti Loading…
[llama3] Add tied weights support CLA Signed This label is managed by the Meta Open Source bot.
#1409 opened Jul 17, 2025 by idoh Draft
[llama4][auxiliary-loss-free load balancing] update expert_bias without backward hooks CLA Signed This label is managed by the Meta Open Source bot.
#1304 opened Jun 16, 2025 by hann-wang Loading…
[PoC] Enable flexible different layout for same mesh via a util function CLA Signed This label is managed by the Meta Open Source bot.
#1550 opened Aug 11, 2025 by fduwjj Loading…
Implement the num_flops_per_token calculation in get_nparams_and_flops() function for Flux CLA Signed This label is managed by the Meta Open Source bot.
#2452 opened Feb 27, 2026 by vivekgoe Loading…
[WIP] Memeory Sharded Tensor (MST) ciflow/8gpu CLA Signed This label is managed by the Meta Open Source bot.
#2378 opened Feb 13, 2026 by weifengpy Draft New Feature, Model, Misc
[PP] microbatch split config CLA Signed This label is managed by the Meta Open Source bot.
#947 opened Mar 7, 2025 by H-Huang Draft
Support finetuning from a pretrained model CLA Signed This label is managed by the Meta Open Source bot.
#1321 opened Jun 20, 2025 by vwxyzjn Loading…
[MoE][PoC] Expert Parallel: dp2ep CLA Signed This label is managed by the Meta Open Source bot.
#732 opened Dec 12, 2024 by tianyu-l Draft
experiments: add nemotron3 model to experiments folder CLA Signed This label is managed by the Meta Open Source bot.
#2187 opened Dec 30, 2025 by aghilann Loading…
4 tasks
feat: add support for DeepEP in Qwen3 parallelization logic CLA Signed This label is managed by the Meta Open Source bot.
#2392 opened Feb 18, 2026 by jordisassoon Loading…
Add PP tracer + DP test CLA Signed This label is managed by the Meta Open Source bot.
#379 opened Jun 1, 2024 by kwen2501 Loading…
Export MoE CLA Signed This label is managed by the Meta Open Source bot.
#1745 opened Sep 23, 2025 by kwen2501 Draft
[DeepSeek] Move seqlen from model config to setup_symm_mem CLA Signed This label is managed by the Meta Open Source bot.
#1017 opened Mar 24, 2025 by kwen2501 Loading…
Add other ops CLA Signed This label is managed by the Meta Open Source bot.
#2033 opened Nov 13, 2025 by wconstab Loading…
ProTip! Filter pull requests by the default branch with base:main.