general family · filed under optimization
Training stability stress testing
A family of tests that amplify relevant training stress within a smaller budget.
Also called stress tests.
- source
- 1
- model
- 1
- 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.
we design a set of stress tests that amplify the relevant stress within a smaller budget
usedevaluation onlyin Qwen3.8-Flash-NextQwen
Filed alongside
Other methods under optimization :: training stability.
Weight paddingActivation or logit clippingAnti-hallucination trainingBehavioral regularizationCross-replica model-weight hash consistency checksElevated constant-learning-rate training stability stress testExponential moving average of checkpointsGradient clippingLearning-rate elevation for training stability stress testingLoss masking for excessively stale tokensOff-policy sample filteringPer-token regularization for off-policy RLSafeguards against training drift and reward hackingSoft droppingWeight clippingWeight decay coupled to learning-rate squared