MatheusMartins

While trying out Ollama, I faced this same issue on models `llama2` and `mistral`. I managed to work around it by using `tinydolphin`, which is a 1.1B Parameter token, replying...