Reading preview
On-policy Distillation with Verifiable Reward:用可验证奖励给同策略蒸馏加上"是非门"
Preparing this article. Please wait.
On-policy Distillation with Verifiable Reward:用可验证奖励给同策略蒸馏加上"是非门"
Select “Open original website” to continue reading.
Preparing this article. Please wait.
Reading preview
Select “Open original website” to continue reading.