Open Models Almanac
Methodsoptimizationlearning-rate schedule

specific method · filed under optimization

No batch-size warmup

Training begins at the target batch size rather than ramping it up during early training.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See learning-rate schedule for the methods filed beside it.

Also called Training without batch-size warmup.

sources
2
model
1
lab adopt it
1
strongest
used

Documented in

Alibaba (Qwen)

Evidence

2 quotes from 2 sources, one card per model, each opening with its strongest.

How sources treat it
used 2
↑