-
Notifications
You must be signed in to change notification settings - Fork 802
Pull requests: pytorch/torchtitan
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
[GraphTrainer] Skip identity-slice rewrite when end is a dynamic Node
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
#3195
opened May 1, 2026 by
aditvenk
Contributor
Loading…
[GraphTrainer] Annotate generated FX code with user source lines
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
#3194
opened May 1, 2026 by
aditvenk
Contributor
Loading…
[MoE] Pad token count to a multiple of sp_size in AllToAllTokenDispatcher
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
#3193
opened May 1, 2026 by
wwwjn
Contributor
Loading…
[MoE] Move shared_experts out of combine() for clean DTensor boundaries
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
#3192
opened May 1, 2026 by
acisseJZhong
Contributor
•
Draft
[MoE] Pad token count to a multiple of sp_size in AllToAllTokenDispater
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
#3191
opened May 1, 2026 by
wwwjn
Contributor
Loading…
[WIP][numerics] Add activation tracer
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
[LoRA] Converter protocol and PEFT external format support
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
[Checkpoint] BaseModel state dict pipeline and pure I/O CheckpointManager
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
Tests HybridEP integration with GraphTrainer
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
#3184
opened Apr 30, 2026 by
syed-ahmed
Collaborator
Loading…
Upgrades torchtitan-ubuntu docker image to 22.04 and CTK 13.0
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
#3183
opened Apr 30, 2026 by
syed-ahmed
Collaborator
Loading…
[Skill] Add fix_ci skill for CI babysitting
CLA Signed
This label is managed by the Meta Open Source bot.
#3179
opened Apr 30, 2026 by
fegin
Contributor
Loading…
[GraphTrainer] Fix FlexAttention precompile bitwise deterministic tests
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
#3178
opened Apr 30, 2026 by
bobrenjc93
Contributor
Loading…
Observability: structured logging + training instrumentation (#3176)
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
fb-exported
meta-exported
#3176
opened Apr 30, 2026 by
felipemello1
Loading…
[WIP] Add bitwise parity test for MoE EP
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
#3172
opened Apr 30, 2026 by
wwwjn
Contributor
Loading…
[WIP]Enable DP-to-EP for MoE inference
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
#3171
opened Apr 30, 2026 by
wwwjn
Contributor
Loading…
[cpu-offloading] Implement prefetching for cpu offloading pass
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
#3166
opened Apr 29, 2026 by
mlazos
Contributor
Loading…
[NOT READY FOR REVIEW][Module][Qwen3-VL] Full config-based sharding + full_dtensor
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
[graph_trainer] Generalize minimal_fx_tracer to module + optimizer roots
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
#3164
opened Apr 29, 2026 by
tugsbayasgalan
Contributor
Loading…
Temporarily disable torchcomms TP+PP+compile test
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
#3163
opened Apr 29, 2026 by
mori360
Contributor
Loading…
[graph_trainer] Re-enable ChunkedCELoss via compute_traceable_grads
ciflow/h100.8
Trigger H100.8 CI
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
#3162
opened Apr 29, 2026 by
tugsbayasgalan
Contributor
Loading…
[NOT READY FOR REVIEW][Full DTensor] Enable full_dtensor for MoE models (qwen3, llama4, gpt_oss, deepseek_v3)
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
[NOT READY FOR REVIEW][Module][MoE] Config-based MoE sharding (qwen3, llama4, gpt_oss, deepseek_v3)
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
[Full DTensor] Config-based Full DTensor for Llama3
ciflow/h100.8
Trigger H100.8 CI
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
#3159
opened Apr 29, 2026 by
fegin
Contributor
Loading…
[graph_trainer] FSDP AG RS overlap
ciflow/h100.8
Trigger H100.8 CI
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
#3156
opened Apr 29, 2026 by
yiming0416
Contributor
Loading…
[will close, do not review] Revert "[Module] Remove LocalMapInnerAttention" and "[rl] Removed pat…
ci-no-td
ciflow/8gpu
CLA Signed
This label is managed by the Meta Open Source bot.
#3154
opened Apr 29, 2026 by
daniellepintz
Contributor
Loading…
Previous Next
ProTip!
Updated in the last three days: updated:>2026-04-27.