On November 2, 2026, four researchers including Rikiya Takehi from MIT CSAIL (Computer Science and Artificial Intelligence ...
Article by Shenchao TechFlowNo training required, no data needed, no gradient descent.Researchers at MIT fed the weights of ...
MIT researchers say GPT-6 Astra edited 0.016% of Qwen3.5-0.8B weights without training, fixing reasoning loops and lifting MATH-500 from 7.9% to 48.1%.
SNACK: truly sparse GPU layers with a custom COO SpMM CUDA kernel — NeurIPS 2026 (arXiv:2610.04093) - JafarBadour/Snack-SNN ...
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