Login

News on Medial

AI language models can exceed PNG and FLAC in lossless compression, says study

ArstechnicaArstechnica · 1y
AI language models can exceed PNG and FLAC in lossless compression, says study

A research paper titled "Language Modeling Is Compression" reveals that DeepMind's large language model, Chinchilla 70B, can achieve better lossless compression on images and audio compared to the PNG and FLAC algorithms. The study suggests that language models like Chinchilla not only excel in text prediction but also in compressing other types of data. This brings forth the idea that effective compression is a form of general intelligence, as it involves identifying patterns and making sense of complexity. The relationship between prediction and compression is bidirectional, allowing compression algorithms to generate new data based on what they have learned. While the paper is not peer-reviewed, it highlights the potential applications of large language models in compression tasks.

View Source

Comments

More Like this

AI fails to detect depression signs in social media posts by Black Americans, study finds

AI fails to detect depression signs in social media posts by Black Americans, study finds

This Week in AI: Addressing racism in AI image generators

This Week in AI: Addressing racism in AI image generators

AI poisoning could turn open models into destructive “sleeper agents,” says Anthropic

AI poisoning could turn open models into destructive “sleeper agents,” says Anthropic

Soket AI Partners Google Cloud To Launch Multilingual AI Model

Soket AI Partners Google Cloud To Launch Multilingual AI Model

AI fails to detect depression signs in social media posts by Black Americans: study

AI fails to detect depression signs in social media posts by Black Americans: study

Generating AI images uses as much energy as charging your phone, study finds

Generating AI images uses as much energy as charging your phone, study finds

Researchers upend AI status quo by eliminating matrix multiplication in LLMs

Researchers upend AI status quo by eliminating matrix multiplication in LLMs

AI-generated propaganda as effective as original propaganda, finds new study

AI-generated propaganda as effective as original propaganda, finds new study

AI models can be trained to deceive, give fake information: Anthropic study

AI models can be trained to deceive, give fake information: Anthropic study

WTF is AI?

WTF is AI?

Download the medial app to read full posts, comements and news.