Bee - Coding Advanced
Bee - Coding Advanced
Advanced
Project Goals
Your mission: Beat the state of the art model (SOTA) by providing your own prompt(s) and response(s) that are:
- Highly creative
In order to produce high quality, SOTA beating tasks, please remember to do the following for every single task:
- Review the Task Type: In the UI there will be a note indicating the task type (e.g., RAG, Test Generation, Bug Fixing, Solution Reasoning).
- Use Perfect Spelling, Grammar, and Punctuation: In order to beat SOTA, you will need to proofread your work and eliminate any grammatical,
- Consider the User: Would the response you created delight the user relative to the prompt? Does it address the prompt completely? Does it
Retrieval Prompt: Problem statement that includes a URL to a text resource (document or web page )
Augmented Response: code and/or natural language explanation that uses the resource mentioned in
Generation (RAG) the prompt.
Test Generation Prompt: Problem statement and code that includes an ask for unit tests.
Response: Code with unit tests and brief explanation of what they are testing
Bug Fixing Prompt: Code that contains a bug and an error message that you get when running the
code. Ask the model to analyze the code and fix the bug.
Response: Fixed code along with analysis of the code and bug, explanation of how it was
fixed.
Solution Reasoning Prompt: Problem statements along with code that solves the problem.
Response: Explanation using bullet points of how the code solves the problem. Analyze
efficiency and correctness of the code too.
Attempting a Task : Step 1 : Follow the Subcategory
Look for the subcategory of the task, and make sure you understand what is
required.
Attempting a Task : Step 2 : Writing The Prompt
A perfect response satisfies every ask within the prompt, adheres to best
practices, and is easy to read.
1. Satisfy the prompt: Make sure that everything that the prompt asks for is
provided by the response, code/explanations of code, etc.
2. Best practices: Code comments, thorough error handling, validation of user
inputs, modular code and efficient code using Swift’s built-in features like
Structs and Enums. Providing example usage of the code helps too.
3. Easy to read: Concise explanations of code. No redundancy in the text that
surrounds the code. Proper use of markdown in the text.
Important Things To Consider 1:
1. LLMs: Use of LLMs to assist your prompt or response writing is strictly
prohibited. Your work will be checked for LLM usage, which results in removal
from the project if found to be using LLMs.
2. Swift Libraries: Try to keep the Swift libraries used in your code as basic as
possible, using only cross-platform libraries if needed. See our list of
accepted/rejected libraries.
3. Grammar/Spelling: Your prompts and responses should not contain any
grammar or spelling errors. This will result in your task being marked as
subpar. Please always check your grammar and spelling will Quillbot.
Important Things To Consider 2:
4. Use of Markdown: Please always use proper markdown for code snippets and
code elements.
```swift
```
5. Mention Swift in your prompts: For any prompt that doesn’t include code,
always mention Swift in the prompt so the model knows what language to work
with.
Important Things To Consider 3:
6. Don’t forget to provide a clear ‘ask’ of the model:
RAG: I need the documents pertaining to Mercury summarized …
Test Generation: Provide me with tests that validate the function …
Bug Fixing: Analyze the code and provide me with the fixed code …
Solution Reasoning: Explain how this code solves the problem and
comment on it’s correctness and efficiency …
7. The Code must run: Make sure the code runs successfully without errors or
warnings. You can use your own IDE or online compilers to test the code.
Retrieval Augmented Generation (RAG) Example 1
FOR REFERENCE ONLY - DO NOT USE THIS PROMPT IN YOUR TASK (your prompts should be
100% created by you!)
Example Prompt 1: I want to start a server-side project with Swift. For this project, I want to
use Vapor and Fluent, so I need to install them. I'm stuck with the installation because I want to
dockerize it so I don't mess up my local machine. I need you to explain to me which are the steps
to install Vapor and Fluent according to this documentation (https://docs.vapor.codes/) and provide
me the commands to do it, please give me the easiest way to set up my Vapor project in Swift.
Response:
https://docs.google.com/document/d/17yhbMdc6uL1Ze4wyqzOCp3ajCYS5kGO-sJt8ZqCM7OY/edi
t?usp=sharing
Retrieval Augmented Generation (RAG) Example 2
FOR REFERENCE ONLY - DO NOT USE THIS PROMPT IN YOUR TASK (your prompts should be
100% created by you!)
Example Prompt 2: I am working on the Bioinformatics course, and I want to create a console
application in Swift, with the information from this website,
https://en.wikipedia.org/wiki/Messenger_RNA. Specifically, I want the information on the
Degradation mechanisms to be summarized, and saved in a dictionary, and the application to give
me the option to choose and view a summary by a key name. It should also allow me to add more
summaries if needed.
Response:
https://docs.google.com/document/d/1IH4jHeBSr0M7kNWl6hWl-8PCgpGujOGeUcYs1pHUx0E/edi
t?usp=sharing
Test Generation Example 1
FOR REFERENCE ONLY - DO NOT USE THIS PROMPT IN YOUR TASK (your prompts should
be 100% created by you!)
Example Prompt 1: I've recently become interested in investing. I've been watching the stock market,
and I'd like to determine what kind of return I should expect for stocks that are prospective investments.
I've created a Swift function that should be able to help me determine the expected rate of return for
various stocks. The function takes the stock ticker and an array of historical annual returns. I want to
simply test that the function is robust by appropriately handling invalid inputs without using any testing
libraries. The test outputs should be easy for any user to read and understand, perhaps by using colored
letters. Invalid inputs to test for include stock tickers that are not between one and four letters in length,
contain special characters or numbers, invalid historical returns, and empty values.
CODE HERE
FOR REFERENCE ONLY - DO NOT USE THIS PROMPT IN YOUR TASK (your prompts should be 100% created by
you!)
Example Prompt 2: I am building a financial services app for mobile phones. I need to be sure that users enter
valid personal information when opening new accounts. I wrote a function in Swift called saveCustomer that
should perform validation of user input and then save the customer data into an array, which serves as local
temporary storage for the lifetime of the application. The customer data should only be saved when all user input is
valid. The user is inputting their name, income, social, zip, and age. The inputs should be tested for edge cases such
as empty values, spaces only, invalid characters, etc. Please write the tests without the use of testing libraries for the
code below, showing each test result in green on the console when a test is passed:
CODE HERE
FOR REFERENCE ONLY - DO NOT USE THIS PROMPT IN YOUR TASK (your prompts should be
100% created by you!)
FOR REFERENCE ONLY - DO NOT USE THIS PROMPT IN YOUR TASK (your prompts should be
100% created by you!)
https://docs.google.com/document/d/1BYksPv8oNdEng_ErhLWddQZR0QMdx7K2CsueZfV6xyw/e
dit?usp=sharing
Solution Reasoning Example 1
FOR REFERENCE ONLY - DO NOT USE THIS PROMPT IN YOUR TASK (your prompts should be
100% created by you!)
https://docs.google.com/document/d/1bDXUM4brz-p2b3UgFI0AH9AjO9Na27R46k7oTCvGCBk/edi
t?usp=sharing
Solution Reasoning Example 2
FOR REFERENCE ONLY - DO NOT USE THIS PROMPT IN YOUR TASK (your prompts should be
100% created by you!)