Rag Fastembed
RAG pipeline using FastEmbed for embeddings generation
FastEmbed is a lightweight, fast, Python library built for embedding generation, maintained by Qdrant. It is suitable for generating embeddings efficiently and fast on CPU-only machines.
In this notebook, we will use FastEmbed-Haystack integration to generate embeddings for indexing and RAG.
Haystack Useful Sources
Install dependencies
Download contents and create docs
Clean, split and index documents on Qdrant
493
FastEmbed Document Embedder
Here we are initializing the FastEmbed Document Embedder and using it to generate embeddings for the documents.
We are using a small and good model, BAAI/bge-small-en-v1.5 and specifying the parallel parameter to 0 to use all available CPU cores for embedding generation.
⚠️ If you are running this notebook on Google Colab, please note that Google Colab only provides 2 CPU cores, so the embedding generation could be not as fast as it can be on a standard machine.
For more information on FastEmbed-Haystack integration, please refer to the documentation and API reference.
Fetching 9 files: 0%| | 0/9 [00:00<?, ?it/s]
Fetching 9 files: 100%|██████████| 9/9 [00:00<00:00, 36900.04it/s] Calculating embeddings: 100%|██████████| 493/493 [00:35<00:00, 13.73it/s]
500it [00:00, 4262.26it/s]
493
RAG Pipeline using Qwen 2.5 7B
Try the pipeline
Calculating embeddings: 100%|██████████| 1/1 [00:00<00:00, 24.62it/s]
(' Dave Grohl is the founder and lead vocalist of the American rock band Foo '
'Fighters, which he formed in 1994 after the breakup of Nirvana, in which he '
'was the drummer.')