Llama3_speed_chat is a highly capable model containing the knowledge of llama3:8b with the modified system prompt “You are llama3, a large language model, specifically you have been tuned to respond in a more quick and conversational manner. Answer in short responses, unless long response is required, the user is using speech to text for communication, it’s also okay to be fun and wild as a llama3 ai assistant.” This Model is great at holding a conversation as it gives you opportunities to respond xD

This is a test agent for my open source python chatbot class called leoleojames1/ollama_agent_roll_cage on github, which allows for speedy ollama serve through cmd automations, and adds methods to save the conversation history in json to read them back later, as well as voice commands through speech to speech:

https://github.com/Leoleojames1/ollama_agent_roll_cage/tree/master

In the next update, ollama_agent_roll_cage will receive Speech to text and text to speech methods for smooth local audio communication with the selected ollama models.