SimeonHristov99/thesis.prethesis.project

Prethesis project used as a baseline for the thesis project.

Prethesis Project

Prethesis project used as a baseline for the thesis project. More information is available here.

Plan of Attack

• Create an initial Plan of Attack.
• Place papers/articles/notes/etc. in the 00_Data folder.
• Create an initial outline in the documentation using task.pdf.
• Extend plan by breaking down the writing process into subtasks.
• Update structure of documentation so that it is ready for review.
• Add links to the papers in MLA Style.
• Get approval for structure.
  • Is the introduction ok?
• Write the implementation part.
  • used Python scripts
  • libraries and what they do
  • explain the pipeline (maybe also include a diagram)
    • explain each step in the pipeline (maybe include commands to start parts of it)
• Write the analysis part.
  • Provide statistics about the final dataset.
    • Explain each sheet in the Excel file.
• Write about train/test split.
  • Maybe also about mini-set creation. In the end, decided not to.
• Write about obtaining Bard baseline.
  • Explain the API (various problems there are when interacting with it).
  • What prompt was used.
  • Comparison with previous backend (was it PaLM - research) and Gemini.
  • Performance and Error analysis on 1,400 samples.

Resources

• The Secrets To Cal Newport's Article Writing System
  • Takeaways:
    1. Read articles, papers, books, etc and place links to them in a text file. For the chosen topic answer questions such as: "What seems interesting here and what doesn't?", "What are the open questions so far?", "What is the context surrounding the topic are?".
    2. Create a folder with every article, paper, book-snippet, etc. that you plan to use. It will set the research scene for what you're writing.
    3. Create an initial outline for the final document. What sections would it comprise of?
    4. More research to support the written outline. Your goal is to know something about the field before you can write a good outline, but you need to have an outline before you understand what else you need to know about the field.
    5. Make a separate directory for each point in the outline and another directory for anything that does not fit into any of the points. For a big writing piece it is good to have at least 30 papers and/or articles.
    6. Commit to writing or doing research every day for a minimum of 4 hours.
    7. As you write when you come across something that you are not sure about / don't know enough about, consult resources from "5.".
• Robert Greene's Proven System For Writing Like A Pro
  • Takeaways:
    1. It has to come from you. You have to have genuine interest in the topic you're writing about. But you also have to be sure that your audience will also find your topic and ideas interesting.
    2. You have to have an element of marketing and publicity. How are other people going to react? Would this resonate with them? Are you clear enough on what your ideas are?
    3. Have a good structure. Don't get lost in your ideas.
    4. Use flashcards to write down your idea so you can structure them more easily. Put a theme, title. Then you can begin to see the big picture and structure the content.
    5. Remain open to changing your ideas and beliefs are you research and write.
    6. Be in an environment that promotes focus, clarity, and a state of flow.
• What's Your Process for Writing New Yorker Articles?
  • Takeaways:
    1. Create an outline of the article.
    2. Breaking the ceil. Do a happy hour writing session. At the end of a work day relatively early in the week, when you feel relaxed try to get started. Pull more sources and ideally write the first paragraph.
    3. Do a morning writing session on Thursday and/or Friday. Try to get half of the article done.
    4. Sunday morning session. Try to finish the article.
    5. Polishing session on the following Monday or Tuesday morning.