miles

mirror of https://github.com/radixark/miles.git synced 2026-10-02 07:14:53 +08:00

Files

T

Tom Chen 3c2f9e5046 Refuse an optimizer a state reset cannot check before resetting it

Resolves op27-30 #1
Resolves op27-17 #3

A take-over that finds no checkpoint zeroes the scheduler and resets the
optimizer, and only then checks the state it rebuilt against the moments adam
keeps. A Muon member rebuilds a `momentum_buffer`, so the check refused it
after the trainer had already been mutated, and the surviving trainer was left
half changed. Refuse the optimizer among the other preconditions, before
anything is touched.

2026-09-03 20:35:32 +08:00

Sort the CI pytest arguments by directory

2026-09-02 18:44:32 +08:00

e2e

Say which fault tolerance scenarios delete their dumps

2026-09-02 23:46:29 +08:00

fast

Refuse an optimizer a state reset cannot check before resetting it

2026-09-03 20:35:32 +08:00

fast-gpu

Forbid positional arguments when constructing ray actors

2026-09-01 18:22:27 +08:00