Tag
A speculative discussion questioning why LLMs are not trained to think in an optimized internal language rather than natural language, and whether that could improve efficiency.