Qwen1.5-110B: The First 100B+ Model of the Qwen1.5 Series
Qwen
Alibaba released Qwen1.5-110B, a 110 billion parameter language model that matches Meta's Llama-3-70B performance on benchmark tests. The model achieves 8.88 on MT-Bench and 43.90% win rate on AlpacaEval 2.0, outperforming the previous 72B version of Qwen1.5. Larger model size without significant changes to training recipes demonstrates that scaling parameter count continues to improve both base and chat model capabilities.
Why it matters
GITHUB HUGGING FACE MODELSCOPE DEMO DISCORD Introduction Recently we have witnessed a burst of large-scale models with over 100 billion parameters in the opensource community. These models have demonstrated remarkable performance in both benchmark evaluation and chatbot arena. Today, we release the first 100B+ model of the Qwen1.5 series, Qwen1.5-110B, which achieves comparable performance with Meta-Llama3-70B in the base model evaluation, and outstanding performance in the chat evaluation, including MT-Bench and AlpacaEval 2.