Running Qwen2.5-Coder (0.5B params) locally on a Mac Intel Core i5 with CodeGPT and Ollama 🤩 In seconds, it explains Python code... all private and secure! Download CodeGPT from the VSCode Marketplace! 👏 Qwen ollama CodeGPT

Check the full article 👇

@Alibaba_Qwen @ollama @codegptAI I recently developed a local swarm powered by Qwen 2.5 3B LLM and was quite shocked by its ability to return structured outputs and function calling. Yet to try out the half billion model. These SLMs look quite promising. 🔗

@Alibaba_Qwen @ollama @codegptAI How does it compare to higher models Qwen models like 3b, in terms of coding

Here’s a comparison of @Alibaba_Qwen models across coding benchmarks. At CodeGPT, we focus on the ‘Fill-in-the-Middle’ metric since it reflects how well a model completes code As the chart shows, larger models like Qwen 3B and up perform better on this metric, with the 32B model leading in accuracy. While bigger models handle more complex coding tasks, they also require more resources, so it’s all about finding the right balance 🧐

@Alibaba_Qwen @ollama @codegptAI is 0.5B good enough?

0.5B is perfect for simple tasks For more complex coding challenges, I’ve had excellent results with the 7B model as well it strikes a great balance between performance and capability! However, keep in mind that the more parameters a model has, the more resources it will consume, so it’s important to choose based on your hardware and task requirements.

@Alibaba_Qwen @ollama @codegptAI whatt, nice. I'll check it !!

@Alibaba_Qwen @ollama @codegptAI @Readwise save thread

@Alibaba_Qwen @ollama @codegptAI How accurate is it, compare to 4o or 3.6 Sonnet?

@Alibaba_Qwen @ollama @codegptAI CodeGPT or Github CoPilot? Who wins❓