Open RobinDeSmet opened 3 months ago
Those smaller models do not have enough power to grasp the instructions in the prompt. They are however much faster. If such a model exists that is powerful enough for this task, we should definitely switch! For now stick with: gurubot/llama3-guru-uncensored:latest
.
The inference time is way to high, we should try to use a much smaller model from Ollama: