Analytics DeepLearning.AI

Data Points

Molmo’s impressive open multimodal models: Llama 3.2 adds vision models and small LLMs

Google’s top Gemini models cut prices, boost performance. Microsoft’s new approach to correct hallucinations. OpenAI releases a multilingual training dataset. Chain-of-thought reasoning has limits.

Machine Learning Research

Robot Server: Google’s table tennis robot triumphs over beginners

A robot that plays table tennis beats human beginners and entertains experts.

Business

Hollywood Embraces Video Generation: Lionsgate teams with Runway to develop a custom fine-tuned video model

The AI startup Runway is helping to retool Lionsgate, the producer of blockbuster movie franchises like The Hunger Games and John Wick, for the era of generated video.

Machine Learning Research

More, Better Open Source Options: Alibaba releases Qwen 2.5 models, raising the bar for open weight LLMs

The parade of ever more capable LLMs continues with Qwen 2.5.

Tech & Society

California Restricts Deepfakes: California enacts laws to regulate deepfakes in politics and entertainment

California, a jurisdiction that often influences legislators worldwide, passed a slew of new laws that regulate deepfakes.

Letters

AI Innovations for Learners: How generative AI is changing education

Last week I spoke at Coursera Connect, the company’s annual conference in Las Vegas, where a major topic was AI and education.

The Batch Newsletter

Hollywood Embraces Video Gen, New Restrictions on Deepfakes, More Open Source Models, Robot Server

The Batch AI News and Insights: Last week I spoke at Coursera Connect, the company’s annual conference in Las Vegas, where a major topic was AI and education.

Data Points

Using GPT to debunk conspiracy theories: Nvidia’s new open vision-language models

Kling 1.5 text-to-video model adds 1080p, editing tools. Codeforces restricts AI use in competition. Identifying whalesong using bioacoustic markers. A prize challenge for improving robotics world models.

Data Points

Alibaba’s impressive suite of open models: Mistral cuts prices across its lineup

Runway adds video-to-video and API. Moshi, a new open speech model. LlamaCoder’s open webapp builder alternative. California restricts synthetic actors and election deepfakes.

Letters

Our New Specialization, the Data Engineering Professional Certificate!: Engineers who can build systems to manage data are in high demand. The Data Engineering Professional Certificate will make you job-ready.

Years ago, when I was working at a large tech company, I was responsible for the data warehouse...

The Batch Newsletter

Models Built for Reasoning, High Gear for Llama 3.1, Brains for Warehouse Robots, Stopping LLMs From Plagiarizing

The Batch AI News and Insights: Years ago, when I was working at a large tech company, I was responsible for the data warehouse.

Machine Learning Research

Reducing Memorization in LLMs: A technique that masks tokens in large language models, protecting data privacy

Studies have established that large language models can memorize the text passages they’ve been trained on repeatedly and regurgitate them when prompted in adversarial and, though rarely, in benign ways.