MIT CSAIL researchers say GPT-6 Astra edited 0.016% of Qwen3.5-0.8B weights to fix reasoning loops, lifting MATH-500 from 7.9% to 48.1% without training.
"To improve an AI model's intelligence, you have no choice but to prepare massive amounts of data and GPUs, and run ...
GPT-6 Astra edited Qwen’s weights and raised MATH-500 from 7.9% to 48.1% in author-run tests. How weight-space programming works—and its limits.
I often see programming recommended in rework programs and career aptitude tests for those returning to work after resigning ...
Unlock the transformative power of self-programming with our comprehensive guide on "I 'Program' Myself" — a strategic, ...
In the strongest example of "weight space programming", as the technique is called, Astra’s intervention raised Qwen3.5-0.8B’s MATH-500 score from 7.9% to 48.1% while touching approximately 0.016% of ...
Build an AI code review pipeline on GitHub Actions with Semgrep, reviewdog and a Python model reviewer. 12 steps, tested code ...
The MIT CSAIL team has proposed a "weight-space programming" method that enables GPT-6 Astra to directly read and modify ...
Local AI models handle most quick tasks without ever going online.
Complete logs of the experiments in the blog post 'LLMs can understand and write neural network weights' - ...
MIT has unveiled an innovative, groundbreaking paradigm dubbed "Weight Space Programming", representing a transformative new ...
Astra identifies game policies from raw weight matrices and fixes reasoning loops in Qwen3.5-0.8B by editing 0.016% of parameters, lifting MATH-500 scores from 7.9% to 48.1% without training loops.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results