microsoft
Ollama models
Hi, is there any local Ollama model that can work with this? I tried many of them, and none produced a working query—they always give me an error. However, when I use GPT or Claude, it works perfectly. Is there any additional setup required to make it work?
Thanks,
There won't be any additional setup required --- unfortunately, some smaller models struggle with following the instructions in the prompt and generating code so they fail pretty often. My personal experience on llama3.2, codellama:7b, qwen-coder-2.5:3b isn't that good
Wondering if larger sized codellama or deepseek-coder would work better? I haven't done much extensive test on smaller models.
Community -- maybe we can build up a small benchmark to test out and get some recommendations for custom models?
