And at last, give suggestions.
Each time you get output, you add context, you add storytelling, you add expert-driven insights, you add one thing, you make modifications, you get the ultimate model of what you wrote for that part, and also you ship it again to LLM, you get outcomes. You say, “That is the ultimate model I will use. You did not embody this, otherwise you omitted this, otherwise you did this, so I made X, Y, and Z modifications. You are violating our guardrails. You are not following our directions.” You can provide that suggestions.
The extra suggestions you give, the less errors you’ll make. The best purpose for me is to realize 70% high quality output. We’re not aiming for 100%. I am not loopy. 70 %, that is the purpose. We need to attain 70% of the great output, making fewer and fewer modifications to the output.
Once more, do not automate your content material. I do not suppose that is sensible if you wish to keep high quality. The trick is to create in batches and personalize the output to make it your personal. Nevertheless, keep in mind that these LLMs are in the end machines, so you will want to maintain coaching.
That is how you should utilize your LLM to create nice content material. thanks.
