Support torchrun-style InfiniTrain multi-process launch - #184

Merged

kilinchange merged 8 commits into

Aug 14, 2026

Commits on Aug 11, 2026

Support torchrun-style InfiniTrain multi-process launch
chen2021673
committed
a00e8b9
test: add basic 8-process config group
chen2021673
committed
5cbb4aa
fix: stabilize distributed training resume flow
chen2021673
committed
aa110a3
fix: correct distributed rank and unique ID handling
chen2021673
committed
fb223c8
fix: require a shared run ID for multi-node launch
chen2021673
committed
21dd48a
fix: separate launcher args from model test args
chen2021673
committed
a1e8738
fix: refine multi-process launcher integration
chen2021673
committed
260f13d
fix(scripts): rename multi-process test cases
chen2021673
committed
ab6ac04