Model techniques map
Techniquespost-trainingrollout & RL infrastructure

specific method · filed under post-training

Discarding early-returned short samples

Discards short samples that return early to reduce length bias and avoid overfitting to overly short sequences.

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 support discarding early-returned short samples to smooth the transition into the steady-state length distribution, and prevent the model from overfitting to overly short sequences

usedpost trainingin DeepSeek-V4.1DeepSeek

Filed alongside

Other methods under post-training :: rollout & RL infrastructure.