Huawei chips refine DeepSeek model in major leap for China's AI self-reliance
9 hours ago
- #AI training
- #Huawei chips
- #China self-reliance
- Chinese chipmakers are effective for AI inference but face challenges with AI training, a more complex process.
- AI pre-training involves teaching a model language through data, while post-training involves teaching it to follow instructions, safety rules, and tasks.
- Researchers conducted full-parameter post-training on DeepSeek's 1.6 trillion-parameter model using a cluster of at least 1,000 Huawei chips.
- Domestic computing power was previously limited to inference tasks, but this project enabled models to self-reflect and adjust, increasing computational demands.
- The collaboration involved Huawei, Shenzhen Loop Area Institute, Harbin Institute of Technology, and Shenzhen Research Institute of Big Data, aiming to boost China's AI industry self-reliance.