How much of a difference is there in the answers from a base gpt3 / davinci model vs a finetuned model? (I'm assuming this finetuning isn't just the addition of a few examples in the prompt)
GPT seems to work well on simple queries. What's your experience trying to generate SQL queries with complex logic across multiple tables?
How much of a difference is there in the answers from a base gpt3 / davinci model vs a finetuned model? (I'm assuming this finetuning isn't just the addition of a few examples in the prompt)
GPT seems to work well on simple queries. What's your experience trying to generate SQL queries with complex logic across multiple tables?