Rishabh Agarwal

This article discusses using classification for value functions in deep reinforcement learning.

This paper reviews the benefits of many-shot learning in language models.

SiT enhances agents' ability to generalize in reinforcement learning through symmetry and attention.

Research shows how MBR decoding enhances translation quality in smaller models.

Gemma 2 offers high performance in a compact size for language tasks.

Study reveals cheaper models may produce better training data for reasoning tasks.

This method helps AIs learn through creating and solving challenges.