There are ways to improve the performance of local LLMs with inference time techniques. You can try with optillm - https://github.com/codelion/optillm it is possible to match the performance of larger models on narrow tasks by doing more at inference.
Comments
There are ways to improve the performance of local LLMs with inference time techniques. You can try with optillm - https://github.com/codelion/optillm it is possible to match the performance of larger models on narrow tasks by doing more at inference.