feat: harden distributed execution and add Python 3.14 support - #29

Merged

zengarden merged 21 commits into

mainfrom

Sep 2, 2026

Commits on Aug 26, 2026

fix: align CUDA DDP preparation with Accelerate
zengarden
committed
114759e
fix(checkpoint): 🐛 support PyTorch 2.6+ checkpoint loading
zengarden
committed
9e9ce9c
fix(tests): 🐛 prevent CPU Ray test hangs
zengarden
committed
fd850d3

Commits on Aug 27, 2026

fix(sampler): 🐛 preserve infinite streams across epochs
zengarden
committed
48e059b

Commits on Aug 28, 2026

Commits on Aug 30, 2026

Commits on Aug 31, 2026

Commits on Sep 2, 2026

fix(redis): 🐛 clean up Redis when the wrapped command exits
zengarden
committed
1967e83
fix(redis): 🐛 bound Redis Cluster startup failures
zengarden
committed
c342ad9
docs(sampler): ✏️ clarify infinite sampler resume semantics
zengarden
committed
6734b00
fix(ray): 🐛 preserve runtime ownership
zengarden
committed
4138ba4
fix(resnet): 🐛 respect dataloader worker count
zengarden
committed
86164bf
fix(config): 🐛 support mapping overrides
zengarden
committed
420f41f
fix(ray): 🐛 bound static cluster readiness
zengarden
committed
e2eb164
fix(redis): 🐛 handle cluster operation errors
zengarden
committed
78f3763
fix(redis): 🐛 validate cluster bus ports
zengarden
committed
6820414
chore(tooling): 🤖 disable mypy gate
zengarden
committed
1faefab
chore(repo): 🤖 remove obsolete todo
zengarden
committed
a879194