Comment by phlip9

Interesting thought. Maybe an LLM would build deeper insight with only one training language. On the other hand, the model might overfit with just one language -- maybe multilingual models generalize better?