README.md

---
license: apache-2.0
task_categories:
- text-generation
language:
- en
tags:
- code
size_categories:
- n<1K
---

Purr-Data Patch Source Code Dataset:
- This dataset is designed for training language models to generate source code for Purr-Data patches. It focuses specifically on patches that output a particular message when a "bang" object is clicked.

Dataset Creation:
- The dataset was created with the goal of evaluating the ability of large language models like Google's 2B GEMMA to be fine-tuned for Purr-Data source code generation.

Dataset Characteristics:
- Content: Each data point consists of two parts:

- Instruction: A textual description of the desired Purr-Data patch functionality. This description focuses on the message the patch should output.
      example instruction => "can you make a Purr-Data patch that displays a funny message?
- Response: The corresponding Purr-Data source code that fulfills the given instruction:
      example response =>
  
        #N canvas 761 0 768 809 10;
        #X obj 260 170 bng 15 250 50 0 empty empty empty 17 7 0 10 #fcfcfc #000000 #000000;
        #X msg 334 25 What do you call a fish with no eyes? Fsh!;
        #X obj 427 335 print; #X connect 0 0 1 0;
        #X connect 1 0 2 0;
  
- Focus: The dataset is restricted to examples where the patch functionality centers around printing a specific message on a bang click.

Potential Uses:
This dataset can be used for various purposes, including:

- Fine-tuning large language models like GEMMA for Purr-Data source code generation.
- Research on text-to-code techniques for visual programming languages.
- Development of code generation tools for Purr-Data.

Note: This dataset provides a starting point for training, and can be further expanded to include more complex Purr-Data functionalities beyond basic message printing.