Sparks of Weight-Space Programming: LLMs can natively read & write neural network weights
1 comments · ai
- Comments15
1 comments · ai
Story details from Lobsters ↗ · linked article on rikiyat.github.io.
Sparks of Weight-Space Programming: LLMs can natively read & write neural network weights was first seen on Lobsters · Hottest on Nov 3, 2026. Since then it has appeared in 65 checks on 2 different days. Its best position was #7 of 25, reached on Nov 3, 10:21 UTC. It dropped off the list after Nov 4, 11:21 UTC. The same link also trended on Hacker News, which raises its cross-platform score.
Position on Lobsters · Hottest at every check (lower is better).
Best rank per day (UTC).
Platform metric as reported by Lobsters.
| List | Best | Days | Checks | Last seen |
|---|---|---|---|---|
| Lobsters · Hottest | #7 | 2 | 65 | |
| YHacker News · Front page Astra can read and write weights to improve Qwen | same link | |||
A few top-level comments from Lobsters, shortened. Read the full thread on Lobsters.
[Might not be as big a deal as claimed](https://honnibal.dev/blog/astra-read-write-misleading)
This means none provided by the harness, not no tools ever. Several of the experiments provided tools (python and other things), and the logs of both directions including what astra reported as reasoning summaries are linked on the blog post. internal python calculator use would likely tend to show up in the logs if…
Archived daily, weekly and monthly lists that include this entry.
Not at the moment. It was last on the list on Nov 4, 11:21 UTC, after 2 days of trending.
#7 on Lobsters · Hottest, on Nov 3, 10:21 UTC. Trendoday checks this list every 30 minutes.
It was first seen on Nov 3, 2026 and has been on the list on 2 days (65 checks) since.