AI界好像烙印不如老鍾吧，看看斯坦福人工智能實驗室主任下午剛發的推特

Christopher Manning

 

@chrmanning

·

4h

 

On https://rank.opencompass.org.cn/leaderboard-llm-v2…, many Chinese LLMs—open & closed—score above GPT-3.5 Turbo & Llama-2-70B-Chat: GLM—Zhipu AI, Yi—01AI, Qwen—Alibaba, Erniebot—Baidu, the 20B InternLM2. If the US’s GPU export bans are meant to stop China building great LLMs, they are not—yet—working.