New VS Code Tool: StarCoderEx (AI Code Generator) — Visual Studio Magazine

New VS Code Tool: StarCoderEx (AI Code Generator)

StarCoder, a new open access Large Language Model (LLM) for code generation from ServiceNow and Hugging Face, is now available for Visual Studio Code, positioned as an alternative to GitHub Copilot.

StarCoder is a transformer-based LLM that can generate code from natural language descriptions, a perfect example of the “generative AI” craze popularized by ChatGPT, the AI-powered, sentient-sounding chatbot from Microsoft partner OpenAI ( and creator of Copilot).

Available as a VS Code extension called StarCoderEx, it can be used to generate code from natural language descriptions in the editor or command palette.

It was born from an open scientific collaboration between Hugging Face (machine learning specialist) and ServiceNow (digital workflow company) called BigCode.

While it’s not strictly open source, it’s parked in a GitHub repository, which describes it like this:

StarCoder is a language model (LM) trained on source code and natural language text. His training data incorporates more than 80 different programming languages, as well as text excerpted from GitHub issues and commits, and notebooks.

“The StarCoder model is designed to level the playing field so developers in organizations of all sizes can harness the power of generative AI and maximize the business impact of automation with the right governance, security and compliance protocols said a May 4 press release from ServiceNow. “This new LLM marks the next major milestone in the BigCode project, an ambitious initiative to develop state-of-the-art AI systems for code in an open and responsible way with support from the open AI research community.”

The same day, Hugging Face published a blog post about the project, involving both StarCoder and StarCoderBase LLM. The company trained a model of nearly 15 billion parameters for 1 trillion tokens, refining the StarCoderBase model for 35 billion Python tokens, which resulted in a new model called StarCoder.

“We found that StarCoderBase outperforms existing open code LLMs on popular programming benchmarks and matches or exceeds closed models such as OpenAI’s code-cushman-001 (the original Codex model that powered early versions of GitHub Copilot). With a length of context of over 8,000 tokens, StarCoder models can process more inputs than any other open LLM, enabling a wide range of interesting applications. For example, by prompting StarCoder models with a series of dialogues, we have enabled them to act as a technical assistant. Additionally, templates can be used to autocomplete code, make code changes via statements, and explain a snippet of code in natural language.We take several important steps toward a secure open model release, including an improved PII authoring pipeline, a new attribution tracking tool, and making StarCoder publicly available with an improved version of the OpenRAIL license. The updated license simplifies the process for companies to integrate the model into their products. We believe that with its strong performance, the StarCoder templates will serve as a strong foundation for the community to use and adapt to their own use cases and products.”

Hugging Face has created a StarCoder – Code Completion Playground that allows users to try out the model by entering a natural language description and viewing the generated code, along with a HuggingChat site that allows users to chat with a guided version of the model, only at demonstration purpose.

When asked about StarCoder, the HuggingChat site responded with, “Starcoder is a natural language processing tool built specifically for developers. Its core capabilities include generating code snippets, providing links to documentation, suggesting variable names, etc., tracking user interactions over time.”

The Hugging Face team also conducted an experiment to see if StarCoder could act as a technical assistant in addition to generating code. They created a tech assistant prompt that allowed the model to act as a tech assistant and respond to programming-related prompts, as shown in the graphic above.

“The model was trained on GitHub code,” Hugging Face said. “As well as Not a template of statements and commands such as “Write a function that takes the square root.” they don’t work well. However, using the Tech Assistant prompt can turn him into a capable Tech Assistant.”

The model is licensed under the BigCode OpenRAIL-M v1 license agreement.

As of this writing, the VS Code extension, with the tagline: “Extension for using GitHub Copilot Alternate (StarCoder API) in VSCode,” has been downloaded 1,890 times since its debut last Friday, May 5th . It earned an average 3.0 rating (0-5 scale) from four reviewers.