Generating coherent videos spanning multiple scenes from text descriptions poses unique challenges for AI. While recent progress enables creating short clips, smoothly transitioning across diverse events and maintaining continuity remains difficult. A …

Researchers from Microsoft have just proposed using a psychological assessment tool called the Defining Issues Test (DIT) to evaluate the moral reasoning capabilities of large language models (LLMs) like GPT-3, ChatGPT, etc. The DIT presents moral dile…

Training giant AI models like GPT-3 requires large resources – thousands of GPUs running for months. As a solo researcher without access to that kind of scale, I can't easily reproduce experiments and findings from papers on huge models. But a new …

LLMs are great with words but can't handle visual tasks like understanding images. Teaching them to use visual tools could make them much more capable. A new paper introduces GPT4Tools – a method to efficiently teach existing LLMs to invoke tools f…

TLDR: New training approach enables smaller AI models to achieve state-of-the-art translation performance Large AI models like GPT-3 have good performance on translation tasks, but some smaller models struggle. Researchers from Johns Hopkins and Micros…

As AI models get bigger, training them requires more and more computing power. Researchers are looking for ways to train these large AI models without needing Google-scale resources. A new paper proposes LongLoRA, a fine-tuning approach that can extend…

Edit: FLAC is the tested audio extension, not MP3 I read the new paper from DeepMind so you don't have to. Here are the key highlights: Despite training on text, langauge models compressed images 43% better than PNG, and audio nearly 2x better tha…

I read the Arxiv paper on CulturaX so you don't have to. Here's my highlights: New open dataset called CulturaX contains text data for 167 languages – far more than previous datasets. With over 6 trillion words, it's the largest multilingu…

If you're creating voice-enabled products, I hope this will help you choose which model to use! I read the papers and docs for Bark and Tortoise TTS – two text-to-speech models that seemed pretty similar on the surface but are actually pretty diffe…

I wrote an in-depth article exploring Vicuna as an alternative to competitor LLMs like ChatGPT, Alpaca, and LLaMA for chat applications. I based it off the research data on the LMSYS.org website and the Github repo for the project. Key findings: Vicun…