Skip to content
Nappai Documentation

MistralAI Embeddings

This component uses the MistralAI API to transform your text into numerical data called “embeddings.” Think of it as translating words into a language that computers understand better, allowing Nappai to perform tasks like finding similar texts or organizing information more effectively.

Relationship with MistralAI

This component directly interacts with the MistralAI API to generate embeddings. It uses MistralAI’s powerful language models to create these numerical representations of your text data. You’ll need a MistralAI API key to use this component.

Inputs

  • Model: Choose the MistralAI model to use for creating embeddings. Currently, only “mistral-embed” is available.
  • Max Concurrent Requests: This sets the maximum number of requests sent to MistralAI simultaneously. A higher number might be faster, but too high a number could overload the system. The default is 64.
  • Max Retries: If a request to MistralAI fails, this sets how many times the component will try again. The default is 5.
  • Request Timeout: This sets the maximum time (in seconds) the component will wait for a response from MistralAI before giving up. The default is 120 seconds.
  • API Endpoint: This is the web address of the MistralAI API. Do not change this unless instructed by Nappai support. The default is already set correctly.
  • Credential: This is where you provide your MistralAI API key, which allows Nappai to access and use the MistralAI service.

Outputs

  • Embeddings: This output contains the numerical representations (embeddings) generated by MistralAI. These embeddings are then used by other Nappai components for tasks like searching, clustering, and comparing text data.

Usage Example

Imagine you have a large collection of customer reviews. You can use this component to generate embeddings for each review. Then, you can use another Nappai component (like a vector database) to find reviews that are similar to each other, allowing you to identify common themes or sentiments.

Templates

This component is used in many Nappai workflows. You can find it within templates that involve semantic analysis, text similarity, and advanced search functionalities. Check the Nappai template library for specific examples.

  • Semantic Text Splitter: This component breaks down large texts into smaller, semantically meaningful chunks, which can then be processed by the MistralAI Embeddings component.
  • Couchbase, Upstash, Chroma DB, Weaviate, Vectara, Redis, PGVector, FAISS, Astra DB, Qdrant, Pinecone, MongoDB Atlas, Milvus, Supabase, Cassandra: These are vector databases that store and search the embeddings generated by this component.
  • Text Embedder: Another component that generates embeddings, but may use different models than MistralAI.

Tips and Best Practices

  • Start with the default settings. Adjust the “Max Concurrent Requests,” “Max Retries,” and “Request Timeout” settings only if you experience performance issues or errors.
  • Ensure your MistralAI API key is correctly entered in the “Credential” input.
  • For very large datasets, consider processing your data in batches to avoid exceeding the API’s request limits.

Security Considerations

Protect your MistralAI API key. Do not share it with unauthorized individuals. Nappai employs secure methods to handle API keys, but it’s crucial to maintain good security practices.