Astra can read and write weights to improve Qwen

Re:Astra can read and write weights to improve Qwen by Nevermark on Monday November 02, 2026 @10:18PM (#50347593)

The way I interpete it, is vast amounts of text are to LLMs, in some ways like vision is to us.

They can triangulate efficiently on the meaning from all the text at once.

It is why they are so good at processing ambiguity.

They can absorb an enormous amount of text and correlate the last thing that was said directly the first thing. That is much more like "seeing" than reading, or any other linear process - even though the field of vision is linear.

[ Parent | on HN ]

Re:Astra can read and write weights to improve Qwen by nakedneuron on Tuesday November 03, 2026 @05:06AM (#50350074)

This.

Lady in red dress scene from Matrix comes to mind.

Our visual system deals with TB of video input every second but we are able to filter through it and spot the important bits. We can even with closed eyes make our mind spin up visual scenes and alter them. I think that's quite possibly a useful explanation of what's happening here.

[ Parent | on HN ]