overclocking-llm-reasoning

Extract and manipulate internal progress vectors in reasoning models to accelerate thinking phases while maintaining answer quality, achieving 80%+ improvements in token efficiency.

Independent GitHub snapshot · Automated classification · Original documentation governs