@kellogh it’s using an OpenAI model AFAIK. in my experience it’s produced more relevant output than vanilla gpt-3.5-turbo for PySpark and SQL generation. some use cases definitely work better than others. what type of tasks have you tried it for? curious what the issues were
it annoys me that it doesn’t know about previous messages in the chat so I have to repeat myself
explain error wouldn’t even rephrase info that was plainly visible in the error message, like column names. IIRC it was an AnalysisException in pyspark
To be fair it does do a great job with fixing code
Add comment