The Daily Index — Thursday, October 19, 2023 · TLDR Index

Across today’s newsletters

Today’s Index

TLDR AI

ADEPT RELEASES MULTIMODAL LLM

Adept has released an extremely simple (no image encoder) and very performant (outperforms 13B models) multimodal model. This article outlines how Adept evaluated the model’s performance and some details about the model.

adept.ai · 8 min

TLDR Web Dev

MODERN WEBSITES

This article breaks down one developer’s opinion of why he feels Next.js is the most reliable JavaScript framework.

felixvemmer.com · 18 min

TLDR Design

14–26% DISCOVERABILITY RATE

ChatGPT-4 was tested to determine how well it could audit 12 different web pages for user experience using its UX suggestions compared to those of a qualified UX professional in a study conducted by Baymard, an organization conducting large-scale research on information technology. In the test, the AI platform made 20% accurate UX suggestions, while 80% of the suggestions had false positives.

baymard.com · 5 min

TLDR AI

USE SMALL MODELS TO STUDY LARGE MODELS

Instabilities in training at large scales are hard to reproduce for most researchers. This paper shows how to reproduce instabilities in small models and verifies that many commonly reported fixes work just as well on these small models.

arxiv.org · 25 min

TLDR AI

ZIPFORMER: A NEW TRANSFORMER FOR SPEECH RECOGNITION

The study introduces Zipformer, a revamped transformer for automatic speech recognition (ASR) that's faster, uses less memory, and outperforms the widely-used Conformer.

arxiv.org · 17 min