[KongchangAI]
Person

Matthew Honnable

A well-known author of open-source software libraries, noted for rigorous criticism of news related to AI technologies.

Timeline (last 90 days)

Nov 4

spaCy author Matthew Honnable criticized the blog post as misleading, conflating two things: "reading weights without tools" and "modifying weights in a full programming environment."

Unverified50%
Nov 4

In the methods table, the "no tools, pure raw numbers" setting appears in only three experiments, involving networks ranging from 17,000 to 172,000 weights.

Unverified50%
Nov 4

Matthew Honnable publicly criticized the headlines of AI "brain-reading patch" news as misleading, pointing out that the math benchmark rise from 7.9% to 48.1% was not achieved purely by reading raw weights.

Unverified50%

All Facts (3)

Source Articles