GaudiEmbedding Examples

This folder contains examples showcasing how to use LlamaIndex with Embeddings integration llama_index.embeddings.gaudi.GaudiEmbedding on Intel Gaudi.

Installation

On Intel Gaudi

bash

pip install --upgrade-strategy eager optimum[habana]
pip install llama-index-embeddings-gaudi

List of Examples

Basic Usage Example

The example basic.py shows how to run GaudiEmbedding on Intel Gaudi and conduct embedding tasks such as text and query embedding. Run the example as following:

bash

PT_HPU_LAZY_ACC_PAR_MODE=1 PT_HPU_ENABLE_LAZY_COLLECTIVES=true python basic.py

Graph RAG Example

GrapgRAG combines Graph Analysis and Retrieval Augmented Generation for richly understanding text datasets. With GaudiHuggingFaceEmbeddings and GaudiLLM, you may now run GraphRAG using local LLM on Intel Gaudi. The example graphrag.py shows how to create a knowledge graph from unstructured text and use that graph to retrieve relevant information for generative tasks via graph search. Follow the instructions to run the example:

Starting NEO4J Database Server

docker run --restart always --publish=7474:7474 --publish=7687:7687 --env NEO4J_AUTH=neo4j/<neo4j-server-password> -v $PWD/data:/data -v $PWD/plugins:/plugins --name neo4j-apoc -e NEO4J_apoc_export_file_enabled=true -e NEO4J_apoc_import_file_enabled=true -e NEO4J_apoc_import_file_use__neo4j__config=true -e NEO4JLABS_PLUGINS=\[\"apoc\"\] -e NEO4J_dbms_security_procedures_unrestricted=apoc.\\\* neo4j:5.22.0

Additional Dependencies

bash

# Intel Gaudi Software Version 1.18.0 or later is required.
pip install llama-index-llms-huggingface
pip install llama-index-llms-gaudi
pip install requirements.txt

# Set the following environment variables:
export NEO4J_USERNAME=neo4j
export NEO4J_PASSWORD=<neo4j-server-password> #default: neo4j
export NEO4J_URL=neo4j://<neo4j-server-host-ip>:7687
export NEO4J_DATABASE=neo4j

PT_HPU_LAZY_ACC_PAR_MODE=1 PT_HPU_ENABLE_LAZY_COLLECTIVES=true python graphrag.py