On the one hand, it did better than chatgpt at understanding what i wanted and actually transforming my data

On the other, truncating my dataset halfway through is nearly as worthless as not doing it at all (and i was working with a single file, maybe hundreds of kilobytes)