Skip to content

Pull requests: pytorch/torchtune

New pull request New
94 Open 1,668 Closed
94 Open 1,668 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

[Gemma2] Use nn.SDPA via MultiHeadAttention CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2844 opened Jun 20, 2025 by Jack-Khuu • Draft
6 of 13 tasks
1
Applies self.act_fn (vs. hardcoded F.silu) in GroupedExperts.forward CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2838 opened Jun 20, 2025 by athms Loading…
2 of 13 tasks
3
Add DCP async checkpointing info to docs CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2837 opened Jun 18, 2025 by ankitageorge Loading…
11 tasks
1
Add DCP async checkpointing for lora dpo recipe CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2835 opened Jun 18, 2025 by ankitageorge Loading…
2 of 10 tasks
1
[DO NOT MERGE] fix main CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2832 opened Jun 16, 2025 by joecummings Loading…
1
[DONT MERGE] Debug branch for the Qwen3 + full model compile CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2831 opened Jun 16, 2025 by anijain2305 Loading…
1
skip compiling opt step instead of erroring if opt_in_bwd=True CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2827 opened Jun 13, 2025 by felipemello1 Loading…
1 of 4 tasks
3
raise error if fsdp_cpu_offload + opt_in_bwd CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2826 opened Jun 13, 2025 by felipemello1 Loading…
1 of 4 tasks
1
Fix #2809, modify attention CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2822 opened Jun 12, 2025 by krammnic Loading…
2 of 13 tasks
1
2
[WIP] Qwen3 MoE support CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2820 opened Jun 12, 2025 by intervitens • Draft
6 tasks
6
[RFC] on-the-fly packing CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2819 opened Jun 12, 2025 by felipemello1 Loading…
14
[WIP] Integrate OptimizerInBackward into SFT distributed recipe CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2818 opened Jun 11, 2025 by joecummings • Draft
1
Test alignment of shared methods in recipes CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2807 opened Jun 9, 2025 by Andrei-Aksionov • Draft
1
Integrate Muon optimizer (2725) CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2803 opened Jun 8, 2025 by Saurabh750 Loading…
1 of 13 tasks
13
Fix command in config CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2796 opened Jun 6, 2025 by krammnic Loading…
1 of 13 tasks
7
[WIP] Proper tool calling support in the torchtune CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2794 opened Jun 6, 2025 by krammnic Loading…
2 of 13 tasks
9
[RFC] Reward modeling CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2788 opened Jun 5, 2025 by krammnic Loading…
1
[RFC] Iterable Dataset CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2785 opened Jun 4, 2025 by felipemello1 Loading…
46
Enable loss parallel, Ungate FP8 CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2782 opened Jun 3, 2025 by nathan-az Loading…
1 of 13 tasks
51
Ungate FP8 + TP CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2781 opened Jun 3, 2025 by nathan-az • Draft
3 of 13 tasks
1
[NOT FOR REVIEW] Full knowledge distillation recipe TP + FP8 CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2767 opened May 28, 2025 by krammnic • Draft
13 tasks
2
[WIP] DSV3 CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2764 opened May 27, 2025 by SalmanMohammadi • Draft
2 of 7 tasks
2
Add LRScheduler.state_dict() to checkpoints CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2762 opened May 23, 2025 by omkar-334 Loading…
2 of 13 tasks
1
4
[WIP][DEBUG] llama4 debugging CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2756 opened May 21, 2025 by IvanKobzarev Loading…
2
Fixing counting number of batches for accumulation through epoch CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#2745 opened May 17, 2025 by wesbz Loading…
7 of 13 tasks
6
ProTip! Mix and match filters to narrow down what you’re looking for.