DeepSeek, a Chinese AI research lab funded by High-Flyer Capital Management, has launched its AI model, DeepSeek-V3.1-Base, on Hugging Face. This advanced model boasts 685 billion parameters and supports various tensor types, including BF16, F8_E4M3, and F32. However, it is notable that the model is not currently deployed by any inference provider, limiting its immediate applications. Users can download DeepSeek-V3.1-Base, but it lacks an official model card and is only available in the Safetensors format to streamline inference workflows. Enhancements include an extended context window, allowing for improved processing and retention of information, as reported by Bloomberg. Despite these advancements, potential users may need to wait for further deployment options by major platforms. For insights into innovative tech advancements, support independent journalism and stay informed on the latest in artificial intelligence and its applications.
Source link
Share
Read more