Efficiency

A new coding framework lets you pipe your own data into large language models. LlamaIndex streamlines the coding involved in enabling developers to summarize, reason over, and otherwise manipulate data from documents, databases, and apps using models like GPT-4.

Fine-tuning a neural network typically involves retraining every layer on new data. But research shows that networks may perform better when fine-tuning modifies only a subset of layers.

Matrix multiplication is executed so often in deep learning, video games, and scientific computing that even a slight acceleration can save substantial amounts of processing time. New work finds ways to speed up this crucial operation.

Both transformers and reinforcement learning models are notoriously data-hungry. They may be less so when they work together. Vincent Micheli and colleagues at the University of Geneva trained a transformer-based system to simulate Atari games using a small amount of gameplay.

Large datasets often contain overly similar examples that consume training cycles without contributing to learning. A new paper identifies similar training examples, even if they’re not labeled.

The route to improving transformer-based language models like GPT-3 and Gopher, which are trained on immense quantities of text scraped from the web, has been to increase their size. But research shows that, given a processing budget, bigger doesn’t necessarily mean better.

Factories are using AI to warn them when equipment is reaching the breaking point. Services that monitor machinery to predict imminent failure and provide guidance on necessary upkeep are booming, The Wall Street Journal reported.

A new approach alters transformers' self-attention mechanism to balance computational efficiency with performance on vision tasks.

A CLIP model whose weights were the mean of an ensemble of fine-tuned models performed as well as the ensemble and better than its best-performing constituent.

You can reduce your model’s carbon emissions by being choosy about when and where you train it.

Vision models can be improved by training them on several altered versions of the same image and also by encouraging their weights to be close to zero. Recent research showed that both can have adverse effects that may be difficult to detect.

Researchers have shown that it’s possible to train a computer vision model effectively on around 66 percent of the pixels in each training image. New work used 25 percent, saving computation and boosting performance to boot.

Equipment that relies on computer vision while unplugged — mobile phones, drones, satellites, autonomous cars — need power-efficient models. A new architecture set a record for accuracy per computation.

Amazon reported long-term success using machine learning to shrink its environmental footprint. The online retailer developed a system that fuses product descriptions, images, and structured data to decide how an item should be packed for shipping.

Researchers discovered a new way to reduce memory requirements when training large machine learning models. Tim Dettmers and colleagues at University of Washington released 8-bit optimizers that store gradient statistics as 8-bit values, while maintaining the same accuracy.