Sergio Paniego(@SergioPaniego):from the vault: on-policy distillation in TRL got 40x faster with three fixes, a generation buffer, batched teacher requests, and binary-encoded logprobs enough to distill from 100B+ teachers, a 4B student gained 39 AIME25 points learning from a 235B one

Sergio Paniego

@SergioPaniego

Machine Learning Engineer @huggingface 🤗 AI PhD. Technology enables us to be more human. 🏳️‍🌈

加入 July 2011

1.7K 正在關注 6.7K 粉絲

Sergio Paniego@SergioPaniego

2026.08.19 12:25

from the vault: on-policy distillation in TRL got 40x faster with three fixes, a generation buffer, batched teacher requests, and binary-encoded logprobs enough to distill from 100B+ teachers, a 4B student gained 39 AIME25 points learning from a 235B one