silx-ai/Quasar-Preview 模型在 Hugging Face 发布，支持 500 万 Token 上下文
英文摘要
A new language model named Quasar-Preview from silx-ai has been published on Hugging Face. The release claims support for a context length of up to 5 million tokens. No further details about architecture, training, or evaluation are provided in the source. It is described as a preview version.
中文摘要
一个名为 Quasar-Preview 的新语言模型由 silx-ai 在 Hugging Face 发布。该发布声称支持最高 500 万 token 的上下文长度。来源中没有提供关于架构、训练或评估的更多细节。它被描述为一个预览版本。
关键要点
A model from silx-ai called Quasar-Preview is now available on Hugging Face.
来自 silx-ai 的模型 Quasar-Preview 现已在 Hugging Face 上可用。
The model specifies a maximum context length of 5 million tokens.
该模型指定的最大上下文长度为 500 万个 token。
It is a preview release, with no technical details disclosed.
这是一个预览版，没有披露技术细节。