Reddit user tests skill transfer between LLMs without fine-tuning

View on Reddit

First manual results from an experiment transferring procedural skill from a large model to a small model without fine-tuning. Small models already understand syntax but gain additional procedural knowledge from larger counterparts.