i couldnt get tool calling to work right with ollama and setting my context to anything above my total gpu megabyte size causes a good percent of the work to be shared with the cpu which sucks and brings things to a crawl. i could keep troubleshooting and maybe get it it to work.
or i could try to figure out how to use other tools like vllm for running models to get tool calls working, but who knows how much time that might take.
so, ive given in to playing around with a copilot subscription so i can check out a few different models. lets see how it goes. maybe ill use them to fix my issues with ollama 😅
#asknostr