qdrant / docs Goto Github PK

Qdrant project documentation

Home Page: https://qdrant.tech/documentation/

docs's Introduction

THIS REPO IS DEPRECATED

Please edit Qdrant documentation here: https://github.com/qdrant/landing_page/tree/master/qdrant-landing/content/documentation

How to use documentation markup features

To add code for different language variants, use blocks of code in the markdown, one after the other, indicating the language:

    ``` bash
        # here your code
    ```
    
    ``` python
        # here your code
    ```

The copy button appears by itself on any outline code blocks.
To emphasize important information, use <aside role="status">your text here</aside> right in markdown

To emphasize dangerous actions or warn users, use <aside role=”alert”>your text here</aside> right in markdown

docs's People

Contributors

Watchers

docs's Issues

Style: impossible to distinguish some links

In the current documentation, it is hard or impossible to distinguish some links.

In alert banners here:

or:

Or here where I don't find it immediately clear that .pem is a literal and mkcert is a clickable link:

I suggest two changes:

underline links
tweak links colors (different link color in banner, different banner text color, or different link color overall)

Error in collections.md

Line 11 of the above file (https://github.com/qdrant/docs/blob/master/qdrant/v1.1.x/collections.md) says, "Distance metrics used to measure similarities among vectors."

I believe an 'are' might be required after 'Distance metrics', resulting in: 'Distance metrics are used to measure similarities among vectors.'

Add Langchain and GPT Index into integrations

Document search on named vectors

We currently only have docs to create a collection with multiple named vectors, but there is no example in the Search section that uses them. It would be good to add at least one example, such that it is demonstrated how it plays together.
It'd also be good to emphasize that the search can only use one named vector at a time (cannot combine them), to avoid confusions.

Incorrect documentation

I am using the python API version 1.1.1 and I cannot get the name of all the collection with client.list_aliases() as exposed herehere.

I need to use client.get_aliases() instead.

BLAS was removed

Hello.
BLAS was removed in the issue
qdrant/qdrant#82

install section contains information about it here https://qdrant.tech/documentation/install/

Update for v0.5.x

Error in payload.md

Line 282 of the said file says "For marked fields will Qdrant will build an index for the corresponding types of queries."
I believe there is an extra "will" before "Qdrant".

Link is expired in docs/qdrant/v1.0.x/search.md file

Current Behavior

Steps to Reproduce

While I was reading through the documentation, I found that these links were not working or expired, I have attached their screenshots for clarity.

Expected Behavior

Possible Solution

Context (Environment)

Detailed Description

Possible Implementation

Disambiguate the use of "store"

We use the verb "store" quite extensively in our documentation.

It is however used with different semantics which can be confusing.

e.g. https://qdrant.tech/documentation/storage/#vector-storage

In-memory storage - Stores all vectors in RAM, has the highest speed since disk access is required only for persistence.

One way of reading this is that vectors are not persisted on disk, meaning that Qdrant would be an in-memory database.

In reality, vectors are stored on disk and are pre-loaded in RAM at startupfor fast access.

Ideally, it should be clear to the user whether something is persisted on disk and/or loaded in memory.