---
title: "Researchers say LLMs can be good at natively reading and writing neural network weights; GPT-6 Astra edited Qwen3.5-0.8B's weights to make it better at math (Rikiya Takehi)"
type: signal
url: https://frontiernewswire.com/s/researchers-say-llms-can-be-good-at-natively-reading-and-writing-e4f14b26
published: 2026-11-02T20:00:53.000Z
byline: "Source: TECHMEME"
---

# Researchers say LLMs can be good at natively reading and writing neural network weights; GPT-6 Astra edited Qwen3.5-0.8B's weights to make it better at math (Rikiya Takehi)

Source: TECHMEME · Published 2026-11-02 · Canonical: https://frontiernewswire.com/s/researchers-say-llms-can-be-good-at-natively-reading-and-writing-e4f14b26

Rikiya Takehi : Researchers say LLMs can be good at natively reading and writing neural network weights; GPT-6 Astra edited Qwen3.5-0.8B's weights to make it better at math — TL;DR … If you were asked to handcraft a neural network that plays Go, what would you do? If you were handed raw neural network weights …

In this story: [GPT-6 Astra](https://frontiernewswire.com/models/gpt-6-astra)

## Sources

- [Read the original at TECHMEME](https://www.techmeme.com/261102/p36#a261102p36)

## Related on Frontier

- [Sam Altman says it’s better for RAM to be in a data center than on your desk](https://frontiernewswire.com/s/sam-altman-says-its-better-for-ram-to-be-in-a-data-center-than-on-your-ad12f597)

---

[More Latest Signal](https://frontiernewswire.com/) · [Signals](https://frontiernewswire.com/verified-reporting)