Complete logs of the experiments in the blog post 'LLMs can understand and write neural network weights'
Stars, forks and merged pull requests each day or month, lined up with the repository's rank on Trendshift Trending
Nov 3 – Nov 4, 2026
Stars +0 in 2 days
Forks +0 in 2 days
Merged PRs 0 in 2 days
Trendshift Trendingwhere the repository ranked each day it made the top 25
Not in the Trendshift top 25 in this period
| Day | Stars gained | Forks gained | Merged PRs |
|---|---|---|---|
| Tue, Nov 3, 2026 | +0 | +0 | 0 |
| Wed, Nov 4, 2026 | +0 | +0 | 0 |
| Total | +0 | +0 | 0 |
So, if you break it down, nothing fancy (still impressive!). The size / resolution of the network is a giveaway that it should be a simple game from Atari period, and then the enumeration becomes…
@liuliu · x.com
Tried to reprod the "read NN weights directly" prompts with @localcodeapp & DeepSeek 4.1 Flash 1017. Out of the 4 games, it got 3. Unlike @simonw's run, mine got the Freeway, but not Pacman (!!!), so…
@liuliu · x.com
Since open-weights models let you to inspect their reasoning traces, it is pretty obvious that the model realizes that it has limited choices (Snake, Pong, Breakout, Tron, Sokoban, Pacman etc.) from…
@liuliu · x.com
