产品功能
- - 一行代码即可启动对大型语言模型(LLM)应用的追踪,提供执行步骤的可见性,并收集用于评估、few-shot示例和微调的数据。
- - 在线评估器可处理每个接收的追踪,自动标记追踪,比人工标记更高效。
- - 可从追踪中构建数据集,用于评估、微调和提示工程,通过检索索引数据集中的语义相似示例来增强提示。
- - 提供无服务器的大型语言模型(LLM)管道构建器,可快速构建和迭代简单的提示和复杂的大型语言模型链。
- - 完全开源且易于自托管,只需几个命令即可开始使用。
Laminar
Our AI engineering platform can track large language model applications with just one line of code, provide visibility into the execution steps, and collect data for evaluation, examples, and fine-tuning. The online evaluator automatically marks the tracking, with far higher efficiency than manual work. At the same time, the platform supports building datasets from the tracking, enhancing the prompting function and improving the model performance by retrieving similar examples. The serverless large language model pipeline builder helps you quickly build and iterate models. The platform is completely open-source, self-hosting is convenient, an...