1 paper
An Yang, Bowen Yu, Chengyuan Li +25
We introduce Qwen2.5-1M, a series of models that extend the context length to 1 million tokens. Compared to the previous 128K version, the Qwen2.5-1M series have significantly enha…