Machine Learning Research
Seemingly an innocuous form of expression, ASCII art opens a new vector for jailbreak attacks on large language models (LLMs), enabling them to generate outputs that their developers tuned them to avoid producing.
Machine Learning Research
Researchers increasingly fine-tune models on synthetic data, but generated datasets may not be sufficiently diverse. New work used agentic workflows to produce diverse synthetic datasets.
Tech & Society
Online publishers are moving to stop AI developers from training models on their content.
Tech & Society
OpenAI is testing an AI-powered search engine in a bid to compete head-to-head with both Google and its close partner Microsoft Bing.
Tech & Society
Meta raised the bar for large language models with open weights and published details about how it built one that outperforms GPT-4o and Claude 3.5 Sonnet by some measures.
Tech & Society
A slimmed-down version of Open AI’s multimodal flagship packs a low-price punch.
Machine Learning Research
Large language models can produce output that’s convincing but false. Researchers proposed a way to identify such hallucinations.
Tech & Society
The word “open” can mean many things with respect to AI. A new paper outlines the variations and ranks popular models for openness.
Machine Learning Research
Low-rank adaptation (LoRA) reduces memory requirements when fine-tuning large language models, but it isn’t as conducive to pretraining.
Business
Amazon hired most of the staff of agentic-AI specialist Adept AI in a move that echoes Microsoft’s absorption of Inflection in March.
Tech & Society
Claude 3.5 Sonnet lets users work on generated outputs as though they were independent files — a step forward in large language model user interfaces.
Machine Learning Research
The technique of model merging combines separate models into a single, more capable model without further training, but it requires expertise and manual effort. Researchers automated the process.