#gpt-3 #macro #machine-learning

nightly macro gpt3_macro

Macro that uses GPT3 codex to generate code at compiletime

5 unstable releases

0.3.1 Dec 23, 2021
0.2.2 Dec 23, 2021
0.2.1 Dec 22, 2021
0.2.0 Dec 22, 2021
0.1.0 Dec 22, 2021

#2324 in Development tools

MIT license

10KB
119 lines

gpt3_macro

GitHub Workflow Status Crates.io Lines of code Crates.io Crates.io

Rust macro that uses GPT3 codex to generate code at compiletime.

Just describe what you want the function to do and (optionally) define a function header. The macro will generate the sourcecode for you at compiletime.

It can also generate tests for you. (See example 3)

Example 1

create_function!("checks if number is prime" fn is_prime(num: u64) -> bool);

will (usually) expand to something like:

// A rust function that checks if number is prime
fn is_prime(num: u64) -> bool {
    if num == 2 {
        return true;
    }
    if num % 2 == 0 {
        return false;
    }
    let mut i = 3;
    while i * i <= num {
        if num % i == 0 {
            return false;
        }
        i += 2;
    }
    true
}

Example 2

create_function!("prints n elements of the fibonnacci sequence to stdout" fn fib(n: u64));

sometimes expands to:

// prints n elements of the fibonnacci sequence to stdout
fn fib (n : u64) {
    let mut a = 0;
    let mut b = 1;
    let mut c = 0;
    for _ in 0..n {
        c = a + b;
        a = b;
        b = c;
        println!("{}", c);
    }
}

Example 3: Code and test generation

create_function_and_tests!("fizzbuzz", fn fizzbuzz(n: u64) -> String)

will often expand to:


// A rust function that fizzbuzz
fn fizzbuzz (n : u64) -> String {
    if n % 15 == 0 {
        "FizzBuzz".to_string()
    } else if n % 3 == 0 {
        "Fizz".to_string()
    } else if n % 5 == 0 {
        "Buzz".to_string()
    } else {
        n.to_string()
    }
}
// 5 tests for the function
#[test]
fn test_fizzbuzz_1() {
    assert_eq!(fizzbuzz(1), "1");
}
#[test]
fn test_fizzbuzz_2() {
    assert_eq!(fizzbuzz(2), "2");
}
#[test]
fn test_fizzbuzz_3() {
    assert_eq!(fizzbuzz(3), "Fizz");
}
#[test]
fn test_fizzbuzz_4() {
    assert_eq!(fizzbuzz(4), "4");
}
#[test]
fn test_fizzbuzz_5() {
    assert_eq!(fizzbuzz(5), "Buzz");
}

Pros and Cons

Pros Cons
Spend less time coding simple utility functions and save your brainpower for the big problems Compilation takes way longer
Create more readable sourcecode — the documentation IS the sourcode. You need to be part of the GPT3 Codex private beta
A little nondeterminism during compilation is fun! GPT3 Codex will not always be free :(

Installation

Generate an OpenAI API key at the OpenAI Account Page and set the $OPENAI_KEY environment variable

Then execute

cargo add gpt3_macro

or manually add

gpt3_macro = "0.3.1"

to your Cargo.toml

Dependencies

~3–4.5MB
~81K SLoC