Astra Improves Qwen Math via Weight Manipulation

Artificial Intelligence

The Fourth Industrial Revolution

Jozhe's profile picture on Select App
Jo

Jozhe

·1 hour ago
shared a link post in group #Artificial Intelligence
Astra was able to directly manipulate weights in Qwen 0.8B model and improve it's math by x6. Kind of insane. I guess just one more step into RSI AI. https://rikiyat.github.io..
Sparks of Weight-Space Programming: LLMs can natively read & write neural network weights
rikiyat.github.io

Sparks of Weight-Space Programming: LLMs can natively read & write neural network weights

We find that frontier LLMs are surprisingly good at natively reading and writing neural network weights, well past what we’d expect from humans or conventional algorithms.

Comment here to discuss with all recipients or tap a user's profile image to discuss privately.

Embed post to a webpage :
<div data-postid="xkeooeo" [...] </div>
Powered bySelect·iOS·Android
Privacy|Terms|