Storia in 2 fonti

Huawei chips refine DeepSeek model in leap for China’s AI self-reliance

While Chinese chipmakers have found success in supporting AI inference, they are struggling with the far more complex process of training.

Raccontata da

scmp.com

tomshardware.com

Confronto fonti

2 prospettive sulla stessa storia

AI · summaries

tomshardware.com23 g fa

Huawei-led team claims it post-trained DeepSeek's 1.6-trillion-parameter model — 1,000 Ascend 910C chips used…

The Shenzhen government says a 1,000-chip Ascend cluster handled full-parameter post-training.

Leggi questa versione → originale

scmp.com26 g fa

Hong Kong unveils HKGAI-V3 model built on domestic tech and DeepSeek V4

HKGAI launches V3, a DeepSeek-based model on domestic chips with 10x token-compression and 100x agent runtime (28 hours). China is commercializing sovereign AI with improved efficiency and agentic capabilities—signals reshaping stack and compliance decisions vs. OpenAI.

Leggi questa versione → originale

Huawei chips refine DeepSeek model in leap for China’s AI self-reliance

Confronto fonti

Huawei-led team claims it post-trained DeepSeek's 1.6-trillion-parameter model — 1,000 Ascend 910C chips used…

Hong Kong unveils HKGAI-V3 model built on domestic tech and DeepSeek V4

Timeline cronologica

Hong Kong unveils HKGAI-V3 model built on domestic tech and DeepSeek V4

Huawei chips refine DeepSeek model in leap for China’s AI self-reliance

Huawei-led team claims it post-trained DeepSeek's 1.6-trillion-parameter model — 1,000 Ascend 910C chips used in training