mKB is a microservice designed to perform semantic searches on datasets. These datasets are pre-processed, chunked, and indexed by creating vector representations (embeddings) of the documents using the mimik dataset-cli tool.
The API documentation is available HERE.
- Create Datasets: Upload pre-processed JSON datasets to the Knowledge Base.
- List Datasets: Retrieve a list of all available datasets.
- Delete Datasets: Remove datasets by their name.
- Upload Chunks: Add specific chunks to an existing dataset using a pre-processed file.
- Similarity Search: Perform searches across datasets based on a prompt, returning the most relevant content.
- Chunk Retrieval: Retrieve individual chunks by their chunkId for detailed inspection.
- Download the latest .zip file for the most recent release HERE.
- Unzip the package.
- Locate the unzipped .tar file; this artifact is needed for the edge microservice deployment.
- If you are deploying the edge microservice for the first time, please follow the edge microservice deployment quick start guide.
- For additional information on ai tutorials, please visit mimik ai quickstart tutorials