Lessons From Red Teaming 100 Generative AI Products

Microsoft Secure Future Initiative september 2024 progress report

SpreadsheetLLM: Encoding Spreadsheets for Large Language Models

MS MARCO Web Search: a Large-scale Information-rich Web Dataset with Millions of Real Click Labels

Multi-Head Mixture-of-Experts

Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone

Who's Harry Potter? Approximate Unlearning in LLMs

DecodingTrust: A Comprehensive Assessment of Trustworthiness in GPT Models

ZeRO++: Extremely Efficient Collective Communication for Giant Model Training - Microsoft Research

LLaVA-Med: Training a Large Language-and-Vision Assistant for Biomedicine in One Day

Reprompting: Automated Chain-of-Thought Prompt Inference Through Gibbs Sampling

Automatic Prompt Optimization with "Gradient Descent" and Beam Search

Imitating Human Behaviour with Diffusion Models - Microsoft Research

Personalized Reward Learning with Interaction-Grounded Learning (IGL) - Microsoft Research

NaturalSpeech 2: Latent Diffusion Models are Natural and Zero-Shot Speech and Singing Synthesizers

Low-code LLM: Visual Programming over LLMs

Inference with Reference: Lossless Acceleration of Large Language Models

Sparks of Artificial General Intelligence: Early experiments with GPT-4

Visual ChatGPT: Talking, Drawing and Editing with Visual Foundation Models

Chat gpt robotics

ChatGPT for Robotics

Neural Codec Language Models are Zero-Shot Text to Speech Synthesizers