Newest Papers
Reinforcement Pre-Training
Qingxiu Dong, Li Dong, Yao Tang, Tianzhu Ye, Yutao Sun, Zhifang Sui, Furu Wei
arXiv·2025
In this work, we introduce Reinforcement Pre-Training (RPT) as a new scaling
paradigm for large language models and reinforcement learning (RL).
Specifically, we reframe next-token prediction as a reasoning task trained
using RL, where it receives verifiable rewards for correctly predicting the
next token for a given context. RPT offers a scalable method to leverage vast
amounts of text data for general-purpose RL, rather than relying on
domain-specific annotated answers. By incentivizing the capability of
next-token reasoning, RPT significantly improves the language modeling accuracy
of predicting the next tokens. Moreover, RPT provides a strong pre-trained
foundation for further reinforcement fine-tuning. The scaling curves show that
increased training compute consistently improves the next-token prediction
accuracy. The results position RPT as an effective and promising scaling
paradigm to advance language model pre-training.
No ratings yet
View paper →Why do we associate the word 'martyr' with suicide?
Substack
Substack·2025
a look at the etymological root of 'martyrdom'
No ratings yet
View paper →Why I Am So Mean - Lyman Stone
Lyman Stone
Substack·2025
Somebody has to clean up the riffraff in this joint.
"But style is just one part of good writing. The other is that it in fact be good, that is, that it fulfill the telos of effortful public communication: to be extremely information dense. Good writing, in the sense of virtuous writing, by moral necessity must make for rather hard reading. Easy reading offends the soul and degrades the man. Good writing should confront the reader with recurrent challenges."
No ratings yet
View paper →I've Been Filling Notebooks for Decades. Here's Why I'm Finally Sharing Them
Substack
Substack·2025
Since my teens, I’ve kept notebooks about all the things that fascinate me, from human nature to the nature of the universe, from the origins of life to the future of life on our planet.
No ratings yet
View paper →What Evolution Can Teach Us About Aligning with AI
Substack
Substack·2025
How the evolutionary logic of fitness interdependence sheds light on the problem of AI alignment
No ratings yet
View paper →