Skip to content

feat(training_engine): add prepare_fwd_bwd_input and wait_for_device to overlap the next step's input copy (-7.2% s/step, 397B RL trainer bench, 128 v7x) - #5718

Merged
khatwanimohit merged 6 commits into
atwigg/mlperffrom
parambole/engine-stage-before-update
Oct 15, 2026
Merged

feat(training_engine): add prepare_fwd_bwd_input and wait_for_device to overlap the next step's input copy (-7.2% s/step, 397B RL trainer bench, 128 v7x)#5718
khatwanimohit merged 6 commits into
atwigg/mlperffrom
parambole/engine-stage-before-update

[MaxTextEngine] Document that staging leaves a pending overflow check…

c79bcfd
Select commit
Loading
Failed to load commit list.
Google CLA / cla/google succeeded Oct 14, 2026 in 8s

✅ All contributors are covered under a CLA with Google

See https://cla.developers.google.com/ for more info about Google's Contributor License Agreement (CLA).

ℹ️ Googlers: Go here to view more details and manage scans for this pull request.

Details

The following contributors were found for this pull request:

✅ c79bcfd Author: @parambole <pa*****le​@google.com>

(Only the first commit for a unique contributor is listed.)

View more details on Google CLA