DeepSeek: Putting a Novel Long-Text Model Architecture Supporting 1M Context to the Test
11 hour ago / Read about 0 minute
Author:小编   

On February 13, DeepSeek's official assistant disclosed in the developer communication group that both its website and APP are currently in the testing phase for a new long-text model architecture. This innovative architecture is designed to support a context length of up to 1M, whereas the API service continues to operate on version V3.2, which is limited to a 128K context.