Tech & Society
Audio Generation Clear of Copyrights: Stability AI releases enhanced text-to-audio generator Stable Audio Open
Sonically minded developers gained a high-profile text-to-audio generator.
Machine Learning Research
Machine Learning Research
Better Teachers Make Better Students: Microsoft‘s Orca 2 strengthens the native reasoning abilities of smaller models
A relatively small student LLM that learns to mimic a larger teacher model can perform nearly as well as the teacher while using much less computation. It can come even closer if the teacher also teaches reasoning techniques.
Machine Learning Research
Richer Context for RAG: RAPTOR, a recursive summarizer, captures more relevant context for LLM inputs
Text excerpts used in retrieval augmented generation (RAG) tend to be short. Researchers used summarization to pack more relevant context into the same amount of text.
Machine Learning Research
Interpreting Image Edit Instructions: Meta’s Emu Edit improves text-to-image generation with task classification.
The latest text-to-image generators can alter images in response to a text prompt, but their outputs often don’t accurately reflect the text. They do better if, in addition to a prompt, they’re told the general type of alteration they’re expected to make.
Machine Learning Research
Brain-Controlled Robots Get More Versatile: NOIR, a system to control robots via electroencephalogram for everyday tasks
Brain-to-computer interfaces that enable users to control robots with their thoughts typically execute a single type of task such as reaching and grasping. Researchers designed a system that responds to a variety of intentions.
Machine Learning Research
Streamlined Inference: Deja Vu, a method that boosts LLM speed by activating only essential neural parts
It’s not necessary to activate all parts of a large language model to process a given input. Using only the necessary parts saves processing.
Letters
Beware Bad Arguments Against Open Source: Big companies are lobbying governments to limit open source AI. Their shifting arguments betray their self-serving motivations.
Inexpensive token generation and agentic workflows for large language models (LLMs) open up intriguing new possibilities for training LLMs on synthetic data...
Machine Learning Research
Predicting Scientific Discoveries: AI predicts scientific breakthroughs using social graphs
A new AI method directs scientists toward promising avenues of inquiry.
Machine Learning Research
A 3D Mesh From One 2D Image: The combination of video diffusion and Neural Radiance Field (NeRF) can produce a 3D mesh from a single image
Video diffusion provides a new basis for generating 3D models.
Machine Learning Research
Tuning LLMs for Better RAG: Meta’s RA-DIT boosts language model output by optimizing text retrieval
Retrieval-augmented generation (RAG) enables large language models to generate better output by retrieving documents that are relevant to a user’s prompt. Fine-tuning further improves RAG performance.
Machine Learning Research
A Transformer Alternative Emerges: Mamba, a new approach that may outperform transformers
An architectural innovation improves upon transformers — up to 2 billion parameters, at least...
Machine Learning Research
More Factual LLMs: FactTune, a method to fine-tune LLMs for factual accuracy without human feedback
Large language models sometimes generate false statements. New work makes them more likely to produce factual output.