setup

virtualenv -p $PYENV_ROOT/versions/3.12.2/bin/python .py3 
source .py3/bin/activate
pip install markdown2pdf

How to use

• create model(send prompts 1~4 to chatGPT)
• create $essay folder under tmp/essays with 2 txt file
  • content.txt
  • topic.txt
• run make step1 $essay, send part1_input.json to chatGPT, get result and save to part1_output.json
• run make step2 $essay, get tmp/essays/${essay}/diff_description.json as part2 input, check prompts/main/5.txt, send to chatGPT, save result to comments.json
• run make step3 $essay

TODO

[x] generate input function [] update prompt to get score [] [prompt update] Task Analysis => "key question" should be based on type of topic [x] convert output under comments and revised into a markdown output [] be careful with escape in input, like origin text use ", input will appear ", which may make chatGPT consider " is origin text.

Optimize prompt

https://platform.openai.com/docs/guides/prompt-engineering

https://platform.openai.com/docs/guides/prompt-engineering/tactic-specify-the-steps-required-to-complete-a-task

step by step:

Use the following step-by-step instructions to respond to user inputs.

Step 1 - The user will provide you with text in triple quotes. Summarize this text in one sentence with a prefix that says "Summary: ".

Step 2 - Translate the summary from Step 1 into Spanish, with a prefix that says "Translation: ".

temperature number or null Optional Defaults to 1 What sampling temperature to use, between 0 and 2. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic.

We generally recommend altering this or top_p but not both.# ielts_writing_coach