Bot releases are hidden (Show)
neo4j-graphrag-python
- Neo4j GenAI package for Python 0.5.0
Latest Release
Published by alexthomas93 about 1 month ago
https://github.com/neo4j/neo4j-genai-python/blob/main/CHANGELOG.md#050
Minor Changes
- Experimental Knowledge Graph Constructor: Introduces a PDF-to-knowledge-graph constructor in experimental mode. This constructor uses a flexible component/pipeline architecture and includes components such as text splitters, a schema builder, an entity-relation extractor, and a Neo4j database writer.
- Hybrid Retriever Fix: Fixes the hybrid retriever query to ensure proper normalization of scores returned by both vector and full-text indexes.
neo4j-graphrag-python
- Neo4j GenAI package for Python 0.4.0
Published by willtai 2 months ago
https://github.com/neo4j/neo4j-genai-python/blob/main/CHANGELOG.md#040
Minor Changes
- Introduced an optional custom_prompt argument to the Text2CypherRetriever class for enhanced customization capabilities.
- Renamed the first parameter of GraphRAG.search from query to query_text for consistency, while ensuring backward compatibility with deprecation warnings for the query parameter.
neo4j-graphrag-python
- Neo4j GenAI package for Python 0.3.1
Published by willtai 3 months ago
https://github.com/neo4j/neo4j-genai-python/blob/main/CHANGELOG.md#031
Patch Changes
- OpenAIEmbedding fixes: Fixed embedding model initialization and removed unused import to prevent ImportError when sentence_transformers is not installed.
neo4j-graphrag-python
- Neo4j GenAI package for Python 0.3.0
Published by willtai 4 months ago
https://github.com/neo4j/neo4j-genai-python/blob/main/CHANGELOG.md#030
Minor Changes
- GraphRAG Integration: Introducing the GraphRAG object to enable a complete Retrieval-Augmented Generation pipeline. This major feature supports context retrieval, sophisticated prompt formatting, and dynamic answer generation, streamlining the use of RAG techniques within Neo4j.
-
Enhanced Customization Options: Added
PromptTemplateandRagTemplatefor customizable prompt generation. Also, a newresult_formatterargument across all retrievers allows for tailored formatting of retrieval results, offering greater flexibility in how results are presented and utilized. -
VectorRetriever Updates: Enhanced to return metadata such as
nodeLabelsandidwithout direct embedding output, simplifying data handling. -
Developer Experience: Enhanced developer interactions with improved documentation, including a new User Guide that provides comprehensive instructions and tips for getting started and effectively using the
neo4j-genai-pythonpackage.
Package Rankings
Top 34.28% on Pypi.org