Qwen-Coder Llama.cpp

Qwen2.5-Coder, a six-model family of LLMs, boasts enhanced code generation, reasoning, and debugging. Trained on 5.5 trillion tokens, its 32B parameter model rivals GPT-4o, offering versatile capabilities for coding and broader applications. This interactive chat interface allows you to experiment with the Qwen2.5-Coder-0.5B-Instruct and Qwen2.5-Coder-1.5B-Instruct coding models using various prompts and generation parameters. Users can select different model variants (GGUF format), system prompts, and observe generated responses in real-time. Key generation parameters, such as ⁣temperature, max_tokens, top_k and others are exposed below for tuning model behavior.