Model techniques map
Techniquespost-trainingpolicy distillation

general family · filed under post-training

On-Policy Distillation

Distillation based on student-induced states or responses, without evidence specifying a more particular variant.

Also called on-policy distillation (OPD).

sources
7
models
5
labs adopt it
3
strongest
core

How sources treat it

One count per evidence span, weakest treatment to strongest.

used 8core 1

Documented in

Further reading

Picked by hand, not extracted: where to read more, not evidence for anything on this page.

Evidence

9 spans quoted from the sources, strongest treatment first.

post-training pipeline replaces reinforcement learning with On-Policy Distillation from 10+ domain specialists

corepost trainingin DeepSeek-V4DeepSeek

a single unified model is trained through on-policy distillation

usedpost trainingin DeepSeek-V4DeepSeek

we employ multi-teacher On-Policy Distillation (OPD; lu2025onpolicydistillation, minillm) as the primary technique for merging expert capabilities into the final model.

usedpost trainingin DeepSeek-V4DeepSeek

the mixed Reinforcement Learning (RL) stage was entirely replaced by On-Policy Distillation

usedpost trainingin DeepSeek-V4DeepSeek

followed by reinforcement learning (RL) and on-policy distillation (OPD; Gu et al., 2024; Lu and Lab, 2025)

usedunclearin DeepSeek-V4.1-FlashDeepSeek

we introduce a second-stage training procedure based on on-policy distillation (OPD), with the goal of distilling the model’s stronger response capabilities under text inputs into the audio-input setting.

usedpost trainingin Qwen3.5-OmniQwen

unified model consolidation via on-policy distillation

usedpost trainingin DeepSeek-V4-FlashDeepSeek

unified model consolidation via on-policy distillation

usedpost trainingin DeepSeek-V4DeepSeek

on-policy distillation using dense, token-level rewards from teacher logits

usedpost trainingin MiMo-V2-FlashXiaomi

Filed alongside

Other methods under post-training :: policy distillation.