No training needed! Opus 5.5 directly hand-writes Transformer weights, scoring 100% on all 13 types of algorithm tasks
不用训练!Opus 5.5直接手写Transformer权重,13类任务全拿100%
Microsoft Research, together with Johns Hopkins University and other universities, proposes interpretable weight-space programming, letting Claude Opus 5.5 write Transformer weights directly in code without backpropagation or gradient descent, reaching 100% accuracy on hidden test sets across 13 types of algorithm tasks, with each iteration relying only on feedback from a single validation score, averaging about 39 rounds. Comparison within the same family shows that stronger models write smaller networks: Haiku 5.5 has a median of about 100,000 parameters, Sonnet 5.5 about 41,600, and Opus 5.5 only 166. The method is also applied to brain encoding models, where on language fMRI data it found a dataset confounding flaw, and on visual cortex tasks its predictions exceed those of existing interpretable models. However, it is limited on NLP tasks such as sentiment classification, reaching only 72% even with parameters inflated to about 190,000; the paper emphasizes that complex representations still require massive data and gradient descent, and this should not be taken to mean that large models no longer need training. Paper: arXiv 2611.18114.
Comparing Claude models in the same family, the paper finds that stronger models write smaller networks, with Opus 5.5 at a median of only 166 parameters, providing a concrete example for understanding differences in model capabilities.