Latest Search
Quote
| Back Zoom + Zoom - | ||||
|
China Unicom: DeepSeek Series Models Upgraded, Saving 30% of Inference Computation on Avg.
Recommend 16 Positive 17 Negative 10 |
||||
|
||||
|
China Unicom recently announced that, following the open-sourcing of its Uni AI "Adaptive Slow Thinking" chain-of-thought foundation model last month, it has optimized and upgraded the DeepSeek series models with "Adaptive Slow Thinking" enhancements. Such upgrade reduces inference computation by an average of about 30%. The optimized model has been made open-source, marking the first such enhancement of the DeepSeek series chain-of-thought models with "Adaptive Slow Thinking" in the industry. AAStocks Financial News |
||||
Copied to Clipboard
 |
||||
