AI publications

A survey of recent advancements in aligning deep generative models with human preferences across language, speech and vision.

This new framework enhances preference optimization for better AI alignment with human values.

Fine-tuning and aligning chain-of-thought responses in LLMs for safer conversational AI.

Introducing APT, an adversarially pre-trained transformer achieving SOTA on small tabular tasks.

A survey covering attack methods, evaluation, metrics and tools for identifying and mitigating GenAI application vulnerabilities.

A simple technique using refusal tokens to control and calibrate refusal behavior in large language models.

Searching for efficient linear operators with optimal scaling laws leading to the development of the BTT-MoE architecture.

Discovering power-law scaling relationships in large time-series transformer models, analogous to those found in language models.

Standard metrics fail in non-English summarization, prompting a need for more nuanced evaluation frameworks.