Model techniques map
Techniquesoptimizationtraining stability

ambiguous · filed under optimization

Safeguards against training drift and reward hacking

A named set of safeguards against training drift and reward hacking; the evidence does not specify their mechanisms.

source
1
models
2
lab adopt it
1
strongest
used

How sources treat it

One count per evidence span, weakest treatment to strongest.

used 1

Documented in

Evidence

1 span quoted from the sources, strongest treatment first.

safeguards against training drift and reward hacking.

usedoptimizationin MiMo-V2.6

Filed alongside

Other methods under optimization :: training stability.