search

QwenLM / Qwen2.5

Qwen2.5 is the large language model series developed by Qwen team, Alibaba Cloud.
8.9k stars 556 forks source link

qwen1.5-72b支持输入和输出的tokens是多大？ #338

Closed TChengZ closed 5 months ago

TChengZ commented 5 months ago

如题，qwen1.5-72b支持输入和输出的tokens是多大？

WangJianQ-0118 commented 5 months ago

模型的配置文件上有吧，32k

TChengZ commented 5 months ago

模型的配置文件上有吧，32k

这个是输入加输出的总token？

jklj077 commented 5 months ago

The Qwen model constitutes a generative language model featuring only a decoder component, such that from the standpoint of its architectural design, there exists no intrinsic distinction between input and output sequences. In essence, it generates tokens progressively without a preordained division between the two. Hence, the model boasts a single sequence length capacity of up to 32,768 tokens, wherein the user can arbitrarily assign varying lengths to serve either as input or output.