Relevance AI - The ML Platform for Unstructured Data Analysis
🌎 80% of data in the world is unstructured in the form of text, image, audio, videos, and more.
🔥 Use Relevance to unlock the value of your unstructured data:
- ⚡ Quickly analyze unstructured data with pre-trained machine learning models in a few lines of code.
- ✨ Visualize your unstructured data. Text highlights from Named entity recognition, Word cloud from keywords, Bounding box from images.
- 📊 Create charts for both structured and unstructured.
- 🔎 Drilldown with filters and similarity search to explore and find insights.
- 🚀 Share data apps with your team.
Relevance AI also acts as a platform for:
- 🔑 Vectors, storing and querying vectors with flexible vector similarity search, that can be combined with multiple vectors, aggregates and filters.
- 🔮 ML Dataset Evaluation, for debugging dataset labels, model outputs and surfacing edge cases.
🧠 Documentation
| Type | Link |
|---|---|
| Python API | Documentation |
| Python Reference | Documentation |
| Cloud Dashboard | Documentation |
🛠️ Installation
Using pip:
pip install -U relevanceai
Using conda:
conda install -c relevance relevanceai
⏩ Quickstart
Login to relevanceai:
from relevanceai import Client
client = Client()
Prepare your documents for insertion by following the below format:
- Each document should be a dictionary
- Include a field
_idas a primary key, otherwise it's automatically generated - Suffix vector fields with
_vector_
docs = [
{"_id": "1", "example_vector_": [0.1, 0.1, 0.1], "data": "Documentation"},
{"_id": "2", "example_vector_": [0.2, 0.2, 0.2], "data": "Best document!"},
{"_id": "3", "example_vector_": [0.3, 0.3, 0.3], "data": "document example"},
{"_id": "4", "example_vector_": [0.4, 0.4, 0.4], "data": "this is another doc"},
{"_id": "5", "example_vector_": [0.5, 0.5, 0.5], "data": "this is a doc"},
]
Insert data into a dataset
Create a dataset object with the name of the dataset you'd like to use. If it doesn't exist, it'll be created for you.
ds = client.Dataset("quickstart")
ds.insert_documents(docs)
Quick tip! Our Dataset object is compatible with common dataframes methods like
.head(),.shape()and.info().
Perform vector search
query = [
{"vector": [0.2, 0.2, 0.2], "field": "example_vector_"}
]
results = ds.search(
vector_search_query=query,
page_size=3,
)
Learn more about how to flexibly configure your vector search ->
Perform clustering
Generate clusters
clusterop = ds.cluster(vector_fields=["example_vector_"])
clusterop.list_closest()
Generate clusters with sklearn
from sklearn.cluster import AgglomerativeClustering
cluster_model = AgglomerativeClustering()
clusterop = ds.cluster(vector_fields=["example_vector_"], model=cluster_model, alias="agglomerative")
clusterop.list_closest()
Learn more about how to flexibly configure your clustering ->
🧰 Config
The config object contains the adjustable global settings for the SDK. For a description of all the settings, see here.
To view setting options, run the following:
client.config.options
The syntax for selecting an option is section.key. For example, to disable logging, run the following to modify logging.enable_logging:
client.config.set_option('logging.enable_logging', False)
To restore all options to their default, run the following:
Changing the base URL
You can change the base URL as such:
client.base_url = "https://.../latest"
🚧 Development
Getting Started
To get started with development, ensure you have pytest and mypy installed. These will help ensure typechecking and testing.
python -m pip install pytest mypy
Then run testing using:
Don't forget to set your test credentials!
export TEST_PROJECT = xxx
export TEST_API_KEY = xxx
python -m pytest
mypy relevanceai
Set up precommit
pip install precommit
pre-commit install
Metadata
Release files for RelevanceAI-dev 3.2.15.2022.9.28.1.41.50.729944
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| RelevanceAI-dev-3.2.15.2022.9.28.1.41.50.729944.tar.gz | 301.5 kB | Without provenance | Details for RelevanceAI-dev-3.2.15.2022.9.28.1.41.50.729944.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| RelevanceAI_dev-3.2.15.2022.9.28.1.41.50.729944-py3-none-any.whl | Python 3 | none | any | Details for RelevanceAI_dev-3.2.15.2022.9.28.1.41.50.729944-py3-none-any.whl |
Total release size: 728.6 kB
Release files / RelevanceAI-dev-3.2.15.2022.9.28.1.41.50.729944.tar.gz
| Download URL | RelevanceAI-dev-3.2.15.2022.9.28.1.41.50.729944.tar.gz |
|---|---|
| Size | 301.5 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
e30f5e56432f6edd47ae03e9a889c03bb04fe558a5c716e62c4da40ef37ef89a
|
|
BLAKE2b-256 checksum How to use checksums |
fd90bdc07baa912ba689dfaa2dc8446f497ba11682f5d79c159bbb31c656f654
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/4.0.1 CPython/3.10.7
|
Release files / RelevanceAI_dev-3.2.15.2022.9.28.1.41.50.729944-py3-none-any.whl
| Download URL | RelevanceAI_dev-3.2.15.2022.9.28.1.41.50.729944-py3-none-any.whl |
|---|---|
| Size | 427.1 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
07fdfaeae9739fe1a7f93999cc2cffec47f0cffd745a2e5733d23621c0705ce8
|
|
BLAKE2b-256 checksum How to use checksums |
711fced90f74077037dc96951001f6338b1242cbfa6527054ac96c3ea9cdafa4
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/4.0.1 CPython/3.10.7
|
Security for RelevanceAI-dev 3.2.15.2022.9.28.1.41.50.729944
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.