fireworks-ai

Enhancing AI with Fireworks AI and Nomic-Embed-Text-V1.5

Tal Peretz

— 1 min read
Enhancing AI with Fireworks AI and Nomic-Embed-Text-V1.5

Fireworks AI has recently integrated the Nomic-Embed-Text-V1.5 model, a powerful language model developed by Nomic AI. This integration is designed to offer developers superior embedding capabilities at a competitive price.

Nomic-Embed-Text-V1.5 Model

The Nomic-Embed-Text-V1.5 model is notable for its use of Matryoshka Representation Learning, which allows for adjustable embedding sizes without a significant drop in performance. The model supports variable embedding dimensions, such as 768, 512, 256, 128, and 64, enabling developers to balance embedding size and cost.

This model is also multimodal, with the nomic-embed-vision-v1 model aligned to the same embedding space as nomic-embed-text-v1.5, facilitating multimodal embeddings.

Usage and Integration

To use the nomic-embed-text-v1.5 model, include a task instruction prefix in the text prompt, such as search_document: <text here> or search_query: <text here>. The model can be accessed through the Nomic Embedding API or integrated with Fireworks AI, which offers an API-compatible interface similar to OpenAI's embeddings API.

Fireworks AI Integration

To integrate the nomic-embed-text-v1.5 model with Fireworks AI, follow these steps:

  1. Create a Fireworks account and obtain an API key.
  2. Install the necessary integration packages, such as langchain-fireworks for LangChain.

Fireworks supports various embedding models, including nomic-ai/nomic-embed-text-v1.5. You can generate embeddings using the Fireworks API and integrate these with vector stores like Chroma or LangChain's InMemoryVectorStore.

Example Implementation

from langchain_fireworks import FireworksEmbeddings

embeddings = FireworksEmbeddings(
    model="nomic-ai/nomic-embed-text-v1.5",
)

This example shows how to generate embeddings using the nomic-embed-text-v1.5 model through Fireworks, which can be used to index and retrieve documents in a retrieval-augmented generation (RAG) pipeline.

Additional Tools and Libraries

The model integrates with various tools and libraries, such as Chroma, a vector database, to enhance developer productivity and facilitate tasks like similarity searches and document retrieval.

By leveraging the nomic-embed-text-v1.5 model through Fireworks AI, developers can create efficient and scalable embedding solutions for various AI applications.

Read more