Neuroscience, latest Github feature and new ML models

Guess what… Github just made your life easier. If you want to find out how, you should definitely check the second article. In this issue we’ll also explore a bit of neuroscience and the case of Sakana AI.

1: ”The brain’s unique take on algorithms”

Link: The brain’s unique take on algorithms | Nature Communications

Driven by a fervent curiosity at the intersection of neuroscience and theoretical computing, we merge top-down algorithmic approaches with bottom-up physics frameworks to revolutionize neural computing technologies. Passionate about unravelling the brain's computational mysteries, we aim to reshape the landscape of artificial intelligence.

2: “Github’s latest AI tool can automatically fix codes”

Link: GitHub’s latest AI tool can automatically fix code vulnerabilities

GitHub’s newest beta feature is using Copilor and CodeQL to automatically detect and fix vulnerabilities during the coding process. The company promises that the autovix will cover 90% of the alert types in the supported languages (JavaScript, TypeScript, Java, Python).

3: ”Sakana AI evolutionary method of creating new models”

Link: Sakana AI evolutionary method of creating new models

Sakana AI has developed an innovative method to create high-performance generative AI models by combining existing models using evolutionary algorithms. With successful creations like EvoLLM-JP and EvoVLM-JP showcasing superior performance, they aim to surpass the existing models in bot image generation and language tasks.

See you next Thursday!