Model techniques map
Techniquespost-trainingagentic post-training

general family · filed under post-training

Multi-harness training

Jointly optimizing a model across multiple harnesses, treating harness diversity as a training dimension.

Also called Multi-harness reinforcement learning.

source
1
models
2
lab adopt it
1
strongest
used

How sources treat it

One count per evidence span, weakest treatment to strongest.

evaluated 1used 2

Documented in

Evidence

3 spans quoted from the sources, strongest treatment first.

We therefore treat harness diversity as an additional training dimension alongside diversity in tasks and environments.

usedpost trainingin MiMo-V2.6-Pro and MiMo-V2.6-FlashXiaomi

we jointly optimize the model across four mini-harnesses and evaluate it on these harnesses and three additional held-out harnesses.

usedpost trainingin MiMo-V2.6-Distill-Qwen-9BXiaomi

we jointly optimize the model across four mini-harnesses and evaluate it on these harnesses and three additional held-out harnesses.

evaluatedpost trainingin MiMo-V2.6-Distill-Qwen-9B

Filed alongside

Other methods under post-training :: agentic post-training.