I worked on this for a long while already haha but I just switched out llama-70b-versatile for oss-gpt-20b since it got discontinued by groq