Anirudh Goyal

Probe Sampling improves efficiency in testing the safety of language models.

A new method enhances reasoning in language models through effective preference learning.

This article examines the metacognitive skills of LLMs and their impact on math problem solving.

A new optimizer shows promise for fine-tuning pretrained models.

This paper introduces -TCVAE for improved representation learning in machine learning.

Typos can greatly confuse advanced language models and affect their responses.