Open-Assistant SFT-1 12B 模型

这是 Open-Assistant 项目的第一次迭代英语监督微调 (SFT) 模型。它基于一个在 2023 年 3 月 7 日之前通过 https://open-assistant.io/ 人工反馈的 Web 应用程序收集的约 22k 个人助理对话的人类示范，对一个经过 Pythia 12B 微调的模型进行了微调。

模型细节

• 开发者： Open-Assistant Contributors
• 模型类型：基于 Transformer 的语言模型
• 语言：英语
• 微调源自： EleutherAI / pythia-12b-deduped
• 代码： Open-Assistant/model/model_training
• 演示： Continuations for 250 random prompts ( sampling code )
• 许可证：Apache 2.0
• 联系方式： Open-Assistant Discord

提示

使用两个特殊的标记来标记用户和助理的对话开始：<|prompter|>和<|assistant|>。每个对话以<|endoftext|>标记结束。

输入提示例子：

<|prompter|>What is a meme, and what's the history behind this word?<|endoftext|><|assistant|>

输入以<|assistant|>标记结束，表示模型应开始生成助理回复。

生成的续写示例（典型采样率 0.2）：

A visual joke often accompanied by text that spreads rapidly from individual
to individual. The exact origin of the term "meme" is unclear, but it was
first used in the 1990s by Richard Dawkins in his book "The Selfish Gene."
He defined a meme as an idea, behavior, or style that is transmitted from
one person to another within a culture.

Dawkins' use of the term caught on among scholars and journalists, who began
to apply the concept to other types of cultural items such as advertisements,
fashion trends, and jokes. Over time, the term "meme" has come to be used
more generally to describe any social behavior that can be repeated and
altered by individuals. Today, the term is widely recognized and applied in
fields ranging from anthropology to psychology to marketing.<|endoftext|>

限制

请参阅 Pythia 12B 基础模型 here 的限制。

模型在回答数学和编码问题方面表现糟糕。

注意幻觉：输出通常是错误的或具有误导性的。回复可能看起来令人信服（乍看之下），但其中包含完全捏造的错误陈述。

此模型只适用于英语对话。