Open Models Almanac
Methodspost-trainingreward modelling

specific method · filed under post-training

Language-consistency reward

Rewards language consistency in the reasoning trace and answer.

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

Also called a language consistency reward.

sources
3
models
2
labs adopt it
2
strongest
used

Documented in

Aleph Alpha

Evidence

3 quotes from 3 sources, one card per model, each opening with its strongest.

How sources treat it
used 3

Kolibri-1

Aleph Alphaadopted2 sources

we used German environments and included a language consistency reward for both the reasoning trace and the answer.

usedpost trainingin KolibriAleph Alpha
1 more from this model

include a language consistency reward for both the reasoning trace and the answer

usedunclearin KolibriAleph Alpha

DeepSeek-V3.2

DeepSeekadopted1 source

For reasoning and agent tasks, we employ rule-based outcome reward, length penalty, and language consistency reward.

usedtraining objectivein DeepSeek-V3.2DeepSeek
↑