The community members are discussing the difference between using Fastembed and the example in the Qdrant hybrid search documentation. The comments explain that Fastembed is an embedding package that provides sparse embeddings, and the Qdrant example shows how to use Qdrant to generate sparse embeddings, with Fastembed as the default but the ability to customize the embedding model. The community members also note that Fastembed supports many embedding models but not all.
fast embed is just an embedding package that happens to provide sparse embeddings
That notebook is showing how to let QdrantVectorStore generate sparse embeddings for you. By default, it's using fastembed, but you can also customize it (which is useful if you host the models on another server and need to make API calls)
