Archive - The Update of The Gradient

Update #73: Against Language Erasure and Better Long-Context Benchmarking

Researchers and governments develop data sets and technology for local languages, and NVIDIA presents a new, more thorough benchmark for long-context…

Apr 23 •

,

, and

Mini-Update #39: 2024 AI Index Report and Autonomous Driving Detection

Stanford publishes its most comprehensive AI index report and researchers develop new methods of driving lane detection using sparse anchors.

Apr 16 •

and

Update #72: NYC's Unfortunate Chatbot and Mixture-of-Depths

NYC's My City Chatbot makes concerning suggestions to users and DeepMind introduces a novel method for optimizing the resource allocation in transformer…

Apr 9 •

,

, and

Mini-Update #38: OpenAI's Voice Engine and LDM Scaling

OpenAI previews its Voice Engine but does not release it and Latent Diffusion Models (LDMs) show powerful capabilities at small size.

Apr 5 •

and

March 2024

Update #71: Neuralink's First Human Trial and Controlling Language Model Hallucinations

Noland Arbaugh plays chess with his mind, and researchers find training strategies that reduce language model hallucinations.

Mar 26 •

,

, and

Mini-Update #37: xAI's Open Source Grok and Apple's MM1

xAI open sources its LLM Grok and Apple releases MM1 – a series of multimodal LLMs and the lessons learned from training them.

Mar 21 •

and

Mini-Update #36: Claude 3 and 1-bit LLMs

Anthropic releases its most powerful chatbot yet, and researchers release an LLM that uses ternary values for its parameters.

Mar 13 •

and

Update #70: Apple Shutters Autonomous EV Project and Griffin + Hawk Compete with Transformers

Apple terminates Project Titan and redirects resources to generative AI, and researchers propose two RNN-based architectures in the SSM line of work…

Mar 12 •

and

February 2024

Update #69: Gemini Overcompensates for Bias and Missing Details in Sora

Gemini sparks AI culture wars and OpenAI releases a state-of-the-art video generation model.

Feb 27 •

,

, and

Mini-Update #35: Sora and Gemini 1.5 Announcements

OpenAI releases a new text-to-video model, and Google's Gemini 1.5 boasts an extremely large context window.

Feb 25 •

and

Update #68: Whispering Indigenous Languages and Neural Net Training Dynamics

Papa Reo explains issues with Whisper's ability to transcribe the Māori language, and researchers find that neural networks learn statistics of…

Feb 13 •

and

Mini-Update #34: Taylor Swift Deepfakes and Efficient Exploration for LLMs

Explicit AI-generated images of Taylor Swift spread across the internet, and DeepMind researchers propose a method for efficiently and effectively…

Feb 7 •

and

#nojs-banner { position: fixed; bottom: 0; left: 0; padding: 16px 16px 16px 32px; width: 100%; box-sizing: border-box; background: red; color: white; font-family: -apple-system, "Segoe UI", Roboto, Helvetica, Arial, sans-serif, "Apple Color Emoji", "Segoe UI Emoji", "Segoe UI Symbol"; font-size: 13px; line-height: 13px; } #nojs-banner a { color: inherit; text-decoration: underline; } This site requires JavaScript to run correctly. Please turn on JavaScript or unblock scripts