Dataset summary

This repository is dedicated to prompts used to perform in-context learning with starcoder . As a matter of fact, the model is an autoregressive language model that is trained on both code and natural language text. It can be turned into an AI-powered technical assistant by prepending conversations to its 8192-tokens context window.

Format

The prompt is a .txt file which contains multiple conversations between a human and the assistant. Here is the format

-----
Human: <instruction>
Assistant: <answer>

-----

Human: <instruction>
Assistant: <answer>
Human: <instruction>
Assistant: <answer>
.
.
.
-----

Use cases

We want the technical assistant to cover a diverse set of use cases

• Code-to-text :
  • What is the purpose of the following code?<code>
  • What is the bug in the following code?<code>
• Text-to-code :
  • Write/Design/Implement a function to <task>
• Code-to-code :
  • Translate this <code> from <programming language> to <programming language>.
• Text-to-text :
  • What is <technical concept>
• General-purpose Q&A
  • What are you?
  • What is your purpose?

Scope of the work

As a model designed for coding tasks, the user should not expect the model to output relevant answers when prompted with a general-purpose question. When it comes to coding requests, the output of the model should be post-processed before testing them.