@jerryjliu0
The recent @huggingface zephyr-7b-alpha model outperforms ChatLlama 70B ๐ฎ We immediately tested it on @llama_index easy-to-hard tasks ๐งช We found that it is the ONLY open 7B model atm that does well on advanced RAG/agentic tasks ๐ฅ๐ Colab: https://t.co/IzehsZZZz9 https://t.co/yB9xwLAwLx