“Mindblowing! 🤯 A 70B open Meta Llama 3 better than Anthropic Claude 3.5 Sonnet and OpenAI GPT-4o using Reflection-Tuning! In Reflection Tuning, the LLM is trained on synthetic, structured data to learn reasoning and self-correction. 👀” The best …