A sketch-guided multi-objective alignment framework for improving the correctness, efficiency, and algorithmic reasoning of code language models.
-
Updated
Jul 29, 2026 - Python
A sketch-guided multi-objective alignment framework for improving the correctness, efficiency, and algorithmic reasoning of code language models.
Structure-aware code language model prototype with AST features, block graph attention, and Contextual Variable Dropout.
Extended version of EyeMulator (ACL 2026) — improving code language models by mimicking human visual attention. Includes distilled gaze priors, per-token attention annotations, a reference implementation, and the full training/evaluation/analysis pipeline.
Add a description, image, and links to the code-language-models topic page so that developers can more easily learn about it.
To associate your repository with the code-language-models topic, visit your repo's landing page and select "manage topics."