Stability AI introduces StableCode for code generation
August 8 2023
Stability AI has announced its first public release of StableCode, an open large language model (LLM) designed to generate programming language code. The model is available at three levels, including a model for general use cases, an instruction model, and a long-context-window model that supports up to 16,000 tokens. It initially supports development in Python, Go, Java, JavaScript, C, markdown, and C++ programming languages. The data used for StableCode comes from the open-source BigCode project, with Stability AI also applying additional filtering and fine-tuning. The StableCode LLM is based on a transformer neural network and uses the rotary position embedding (RoPE) approach.
Back to Breaking AI News
What does it mean?
- Large Language Model (LLM): A type of artificial intelligence model that is capable of generating human-like text based on the input it's given.
- Instruction model: A type of program that is designed specifically to follow instructions given by a user or another piece of software. It follows a set of rules or protocols to perform tasks.
- Long-context-window model: A type of program that can process and make sense of larger amounts of data or information at once. The "16,000 tokens" refers to the amount of individual pieces of information it can handle in one go.
- Open-source: Something that is open-source is available for anyone to use or modify. It is usually created by a community of developers who share their work freely with others.
- Filtering and fine-tuning: The process of refining or adjusting the data or model to improve performance or to make it more suitable for a specific task.
- Transformer neural network: A type of artificial intelligence model that can understand the context of data it's given, such as the relationships between different words in a sentence. It's often used in tasks that involve processing natural language.
- Rotary position embedding (RoPE) approach: A technique used in transformer models to help them understand the order or position of data, such as the sequence of words in a sentence or the order of events in a list.