docs: Add course content for Day 1 #1930
Conversation
✅ Deploy Preview for condescending-goldwasser-91acf0 ready!
To edit notification comments on pull requests, go to your Netlify project configuration. |
qdrant-landing/content/course/essentials/day-1/movie-search-system.md
Outdated
Show resolved
Hide resolved
| """Semantic chunking: uses embedding similarity to find natural breaks""" | ||
| from llama_index.core import Document | ||
|
|
||
| semantic_splitter = SemanticSplitterNodeParser( |
There was a problem hiding this comment.
Sabrina says in the video that if one wants to learn more about the parameters of this one, we'll provide links below
There're no links afaik
Maybe we should add something?
… clarify chunking constraints
qdrant-landing/content/course/essentials/day-1/pitstop-project.md
Outdated
Show resolved
Hide resolved
…ayload indexes for filtering and grouping to pitstop project.
|
|
||
| ```python | ||
| # Group by movie name to get unique recommendations | ||
| response = client.query_points_groups( |
There was a problem hiding this comment.
Revisiting my comment on "groups" won't even work without an index in other deployments
So I assume it's connected to the strict_mode config
So most probably if it's disabled (or like here, :memory: mode, where we don't have HNSW at all), it still will work without payload_index, just ineffectively (full scans)
Which is ig fine for this lesson, but I'd still keep warning people that the moment they switch to production, payload indices on group fields should be always set up and before building HNSW, if filterable HNSW is needed
There was a problem hiding this comment.
So most probably if it's disabled (or like here, :memory: mode, where we don't have HNSW at all), it still will work without payload_index, just ineffectively (full scans)
I am not sure about the claim that we don't have HNSW when we use :memory: mode. My understanding is that HNSW index is still build but ephemeral and lost on process exit.
There was a problem hiding this comment.
client = QdrantClient(":memory:") # or QdrantClient(path="path/to/db") for local mode and persistent storage
So the parameter for QdrantClient just defines where Qdrant runs and where data is stored, not if HNSW is build:
QdrantClient(":memory:") --> Ephemeral. No Qdrant server.
QdrantClient(path="path/to/db") --> Persistent storage. No Qdrant server.
QdrantClient(url="http://localhost:6333") --> Persistent in server’s storage path.
QdrantClient(url="https://xxx.cloud.qdrant.io:6333", api_key="...") --> Persistent in cloud.
2 participants
No description provided.