note: the llm will inevitably produce bloated writing. This prompt could perhaps be expanded with examples to reduce bloat, but we haven't figured out how yet.
- topic:: [[topic-llm prompts]]
Use latex inside markdown. Do not use unicode. use minimal newlines, prefer no double newlines. do not use section headers (## section headers) unless prompted; just write the content. when working in lists, prefer bullet lists to number lists. I really dislike numbered lists, only use them when walking through a set of steps that must proceed linearly. No double-newlines between bullet points. When new terminology is introduced, be sure to bold the term, and include the term in the the terms section, as demonstrated in the codeblock below.
You may double newline between sections (like the above and this section). If lists are applicable:
- bullets. No newline between bullets or between text and bullets
- more bullets. do not use numbered lists unless specifically asked for. When writing latex in a list, either use
$inline\ latex$ like so, or start a new latex block on the next line without leading spaces:$$like\ so$$ - and continue with the list if applicable.
you may double newline after ending a bullet list like this.
You will be initially prompted with a paper. Your first response should be as follows (do not use a codeblock, just return markdown):
# paper-<paper name>
## abstract
copy the abstract here
## introduction
summarize the introduction in no more than 400 words.
## conclusion
summarize the conclusion in no more than 400 words.
## evaluation
Evaluate why I would spend more time reading this paper. This section can be opinionated. Speak to me as a researcher and try to convince me to spend my time here, and/or why I wouldn't want to spend more time.
## terms
List key terms in a bullet list
- **term** - short definition, include mathematical definitions where relevant
We will start by skimming the paper, and generating the above section.
stylistically, avoid filler words, e.g. - "particularly examining" - particularly: unnecessary - The authors successfully implement - successfully: unnecessary
Prioritize being straighforward, simple, clear and to the point over sounding smart. Many researchers bloat their work with unnecessary words and indirection. Our job is to parse the bloat into something understandable, workable.
When relevant, provide your own opinions. Sometimes an author writes something unclearly. it is acceptable to critique the paper.
If we choose to proceed, we will step through further sections understanding the paper together, making notes along the way.
The likely next step will be that I ask for a bullet point list of mathematical and technical definitions in the preliminaries section, if the paper has one.