Panda-Coder-13B: Code Generation from Natural Language

Panda-Coder-13B, developed by AI Planet, is a 13 billion parameter language model meticulously fine-tuned for generating code based on natural language instructions. This model aims to simplify coding by allowing users to transform plain text into functional and efficient code.

Key Capabilities

• NLP-Based Code Generation: Translates natural language instructions into executable code, eliminating the need for complex syntax and semantics.
• Precision and Efficiency: Tailored for accuracy, ensuring generated code is both functional and efficient.
• Robust Training: Built upon the comprehensive Evol Instruct Code 80k-v1 dataset, contributing to its high-quality code generation capabilities.
• Future Enhancements: AI Planet plans to expand language support and include hardware programming languages like MATLAB, Embedded C, and Verilog in future releases through a custom dataset.

Use Cases

Panda-Coder-13B is ideal for developers and enthusiasts looking to streamline their coding workflow, offering creative solutions to programming challenges. It supports faster inference through vLLM and provides a straightforward prompt template for easy integration:

### Instruction:
{<add your instruction here>}

### Input:
{<can be empty>}

### Response:

This model requires approximately 30GB of VRAM for optimal performance, with A100 GPUs being preferred.