dgtalbug/stable-code-instruct-3b
dgtalbug/stable-code-instruct-3b is a 2.7 billion parameter decoder-only transformer model from Stability AI, instruction-tuned for multi-language code generation and conversational coding assistance. It excels at general-purpose code generation across various programming languages and shows strong performance on MultiPL-E benchmarks. This model is primarily designed as a base for fine-tuning specialized code assistants, offering a solid foundation for domain-specific applications.
Loading preview...
Overview
dgtalbug/stable-code-instruct-3b is a 2.7 billion parameter decoder-only transformer model developed by Stability AI. It is specifically instruction-tuned for multi-language code generation and providing conversational coding assistance. This model serves as an excellent starting point for developers looking to create specialized code assistants, including those fine-tuned with domain-specific datasets.
Key Capabilities
- Multi-language Code Generation: Capable of generating code across various programming languages.
- Conversational Coding: Instruction-tuned to perform well in conversational coding scenarios.
- Strong Benchmarking: Demonstrates solid performance on the MultiPL-E benchmarks, indicating its proficiency in code-related tasks.
Performance Highlights (MultiPL-E pass@1)
- Python: 32.4%
- C++: 30.9%
- Java: 32.1%
- JavaScript: 32.1%
- PHP: 24.2%
- Rust: 23.0%
Good For
- Serving as a base model for further fine-tuning projects.
- Developing specialized code assistants.
- Applications requiring general-purpose code generation and conversational coding support.