Guides
Embedding Models
A comprehensive guide to selecting and using embedding models in your application.
Hymalaia supports integration with several popular embedding model providers. This flexibility allows you to choose the model that best aligns with your performance, language, and efficiency needs.
OpenAI Models
text-embedding-3-small
- Description: Newer, more efficient embedding model
- Strengths: Great balance between performance and efficiency
text-embedding-3-large
- Description: Large embedding model in the OpenAI lineup
- Strengths: Best performance among OpenAI’s offerings
OpenAI Credentials Setup
- Sign up at OpenAI.
- Generate an API key.
- Enter your OpenAI key in the Hymalaia Admin Panel.
Cohere Models
embed-english-v3.0
- Description: Optimized for English-language content
- Strengths: Strong performance for most English use cases
embed-english-light-v3.0
- Description: Lightweight version for faster performance
- Strengths: Efficient for simpler or high-volume tasks
Cohere Credentials Setup
- Create an account on Cohere.
- Follow their documentation to obtain an API key.
- Use the API key in Hymalaia to configure embedding.
Voyage Models
voyage-large-2-instruct
- Description: Instruction-tuned, high-performing embedding model
voyage-light-2-instruct
- Description: Lighter model with a performance/efficiency trade-off
Voyage Credentials Setup
- Sign up at Voyage AI.
- Follow their integration guide to get your API key.
- Enter your Voyage API key in Hymalaia.
Vertex AI (Google)
gecko
- Description: Google’s powerful and efficient embedding model
- Note: Requires Google Cloud Platform (GCP) setup
Vertex AI Credentials Setup
- Create a GCP account.
- Create a new project.
- Enable the Vertex AI API.
- Create a service account with Vertex AI permissions.
- Generate a JSON key for that service account.
- Upload the JSON key in Hymalaia.
Choosing the Right Model
Factor | Recommendation |
---|---|
Task Complexity | Use larger models like text-embedding-3-large or voyage-large-2-instruct |
Language Specificity | Cohere’s models are ideal for English-specific applications |
Performance vs. Efficiency | Consider -light models for faster processing with some trade-offs |
Setup Complexity | OpenAI and Cohere are simpler to set up; Gecko requires GCP configuration |
Best Practices
- Experiment: Try different models with your real data to compare results.
- Monitor: Track effectiveness across different use cases.
- Update Regularly: Check for updates or new model versions from providers.
- Secure Credentials: Use the Hymalaia Admin Panel to securely store API keys or upload credentials.
📌 Note: Always refer to the provider’s official documentation for the most accurate and current information.