A high-performance Gumbel MCTS implementation in pure Python that is 10x faster than vanilla MCTS. CLI tool for search optimization.
-
Updated
Mar 27, 2026 - Python
A high-performance Gumbel MCTS implementation in pure Python that is 10x faster than vanilla MCTS. CLI tool for search optimization.
Implements the "VISion On Request" arXiv paper technique to dynamically select vision-language interactions. It skips 90% of image tokens during VLM inference, measuring the accuracy drop on VQA datasets.
Add a description, image, and links to the research-replication topic page so that developers can more easily learn about it.
To associate your repository with the research-replication topic, visit your repo's landing page and select "manage topics."