Beginners course module 1#2494
Conversation
Module 1-3 pages don't exist yet, so the syllabus links to them were 404ing the lychee internal link checker.
✅ Deploy Preview for condescending-goldwasser-91acf0 ready!
To edit notification comments on pull requests, go to your Netlify project configuration. |
kanungle
left a comment
kanungle
left a comment
There was a problem hiding this comment.
This is a good start. Left some comments. Not sure why Module 0 changes are in this PR
|
|
||
| # Let's Understand Search | ||
|
|
||
| Understand why traditional search fails, how semantic (vector) search solves it. |
There was a problem hiding this comment.
I think this is misleading. I think it should be something like:
Understand why traditional search struggles and how modern semantic search improves it.
| |-------|----------------------|--------| | ||
| | car repair | automobile maintenance guide | ❌ Missed | | ||
| | cheap flights NYC | affordable airfare to New York | ❌ Missed | | ||
| | Apple stock | fruit company disambiguation? | ✔ Match | |
There was a problem hiding this comment.
This line doesn't make sense to me
| | cheap flights NYC | affordable airfare to New York | ❌ Missed | | ||
| | Apple stock | fruit company disambiguation? | ✔ Match | | ||
|
|
||
| ### The Three Core Failure Modes |
|
|
||
| Over time, search systems became more diverse. However, they all shared the same fundamental ceiling: they work on words, not meaning. | ||
|
|
||
| ### Evolution of Search Techniques |
There was a problem hiding this comment.
is this actually a chronological evolution? I didn't look it up
| "Does this document contain the same words?" | ||
|
|
||
| **Semantic search asks:** | ||
| "Does this document mean the same thing?" |
| | Cosine | Text similarity, NLP models | Robust to different vector magnitudes. Most common default. | | ||
| | Dot product | When embeddings are normalized | Faster than cosine if vectors are unit-normalized at index time. | | ||
| | Euclidean (L2) | Image embeddings, spatial data | Sensitive to magnitude - works best with models trained for it. | | ||
|
|
| # SKU-48291 (score: 0.89) ← only correct result returned | ||
| ``` | ||
|
|
||
| ### Key insight |
|
|
||
| - **Dense**: Semantic / vector - Intent, vibe, meaning | ||
| - **Sparse**: BM25 / keyword - Exact terms, rare tokens | ||
| - **Filters (Optional)**: Payload constraints - Hard rules, access control |
There was a problem hiding this comment.
I'm not aware of any of our content that mentions "filtering" as a component of Hybrid Search
| - **Agentic AI systems** - multi-step agents that query different data sources sequentially | ||
| - **E-commerce** - find semantically similar products, then filter by price, brand, and availability | ||
| - **Knowledge bases** - semantic over documents, keyword for exact references and code snippets | ||
|
|
|
|
||
| ## 8. References & Further Reading | ||
|
|
||
| - **Qdrant Documentation** - [Qdrant Overview](https://qdrant.tech/documentation/concepts/) |
There was a problem hiding this comment.
Name it "Qdrant Concepts"
| - **Qdrant Concepts** - [Qdrant Overview](https://qdrant.tech/documentation/concepts/) | ||
| - Overview of Qdrant's vector search engine - collections, points, payloads, and APIs. | ||
|
|
||
| - **Distance Metrics Deep Dive** - [Distance Metrics - Qdrant](https://qdrant.tech/documentation/concepts/#distance-metrics) | ||
| - Cosine, dot product, Euclidean, and Manhattan - when to use each. | ||
|
|
||
| - **Filtering & Hybrid Search** - [Filtering - Qdrant](https://qdrant.tech/documentation/concepts/filtering/) | ||
| - Payload filter syntax, indexed fields, and combining filters with vector queries. | ||
|
|
||
| - **RAG Tutorials** - [RAG Tutorials - Qdrant](https://qdrant.tech/rag) | ||
| - End-to-end retrieval-augmented generation tutorials using Qdrant as the retriever. |
There was a problem hiding this comment.
These links need to be href'd on the text itself, instead of beside it.
Example, just:
[Filtering & Hybrid Search]((https://qdrant.tech/documentation/concepts/filtering/)
| - How similarity really works under the hood - and when it fails. | ||
| - Your first Qdrant collection: points, payloads, and your first query. | ||
|
|
||
| End of Module 1. Continue to Module 2: First Principles of Vector Search. |
There was a problem hiding this comment.
This needs to be a button link, like in the essentials course
There was a problem hiding this comment.
make this change on all modules
There was a problem hiding this comment.
I didn't add it intentionally because it might lead to PR runs failure because of disjoint modules, will add once everything is approved
kanungle
left a comment
kanungle
left a comment
There was a problem hiding this comment.
Progress since last round is real (formula, Manhattan row, concepts naming, intro line all fixed), but several items from my previous reviews are still open, so requesting changes again.
Still open from my last reviews:
- Module 0 files are still in this PR. Please remove
course/beginners/_index.mdand themodule-0/files so this PR only adds module 1; those shared files belong to #2492. Same request going out on every module PR. - Line 202:
- and more...Cut it or name a real use case. - Lines 213–226: reference links. Href the text itself:
[Qdrant Concepts](https://qdrant.tech/documentation/concepts/), drop the link-beside-text pattern for all four. - Title Case on headings throughout: "Today's Path", "Core Limitation", "What You'll Learn", and so on.
- Syllabus still says "Hybrid Search: Dense + Sparse + Filters" in
_index.md. Same point I made on the module body: filters aren't a hybrid-search component. "Dense + Sparse". (Moves to #2492 once the shared files come out of this PR.)
New items:
6. Line 141 is a broken sentence: "A score of 0.0 means on the other hand can be interpreted as…". Suggest: "A score of 1.0 means the vectors point in the same direction. A score near 0.0 means the texts are semantically unrelated." Also drop "(identical meaning)"; cosine 1.0 is same direction, not identical meaning.
7. Line 149: "around 0.847". Did you run this? Three decimals plus "around" don't mix, and the score depends on the model version. Run it and quote the real number, or say "around 0.85".
8. Line 83 table row is mislabeled and the example is broken: typo tolerance is fuzzy matching, not "keyword matching", and "(receive → receive)" is the same word twice. → "Fuzzy matching | Tolerance for typos (recieve → receive)".
9. Line 71: SPLADE doesn't belong in the pre-semantic evolution list. It's a learned sparse model that came after dense retrieval. Drop it here; it's covered properly in Module 3. (This answers my earlier chronology question: no, it wasn't chronological.)
10. Diagrams are off-brand. The colors aren't from the Qdrant color palette and the shapes aren't from our diagram system. Rebuild them from the Docs/Diagrams Figma component library using the Qdrant palette.
11. No exercise. The module is all reading. Add one small task: embed three query/document pairs with the section 4 snippet and compare cosine scores.
12. "Best of both worlds" (line 211) → say what it does: "Covers both exact terms and intent".
13. _index.md: "vector databases" → "vector search engines" (hard rule), and "Duration: Multiple modules" is vague, this course has 6. The Path and syllabus should list all six modules.
Noting for launch, not blocking: the module video placeholder at the top will need the real video before the course goes live.
3 participants
PREVIEW
This PR is still work in progress. Formatting and images will be added soon