https://github.com/eto-ai/rikai-pg

Postgres Rikai Extension
https://github.com/eto-ai/rikai-pg

deep-learning machine-learning postgresql pytorch rikai tensorflow

Last synced: about 2 months ago
JSON representation

Postgres Rikai Extension

• Host: GitHub
• URL: https://github.com/eto-ai/rikai-pg
• Owner: eto-ai
• License: apache-2.0
• Created: 2022-05-10T17:04:10.000Z (over 2 years ago)
• Default Branch: main
• Last Pushed: 2022-06-11T05:09:36.000Z (over 2 years ago)
• Last Synced: 2023-03-03T23:18:53.435Z (almost 2 years ago)
• Topics: deep-learning, machine-learning, postgresql, pytorch, rikai, tensorflow
• Language: Jupyter Notebook
• Homepage:
• Size: 56.6 KB
• Stars: 5
• Watchers: 3
• Forks: 0
• Open Issues: 4
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Postgres Rikai Extension

Status: Proof of concept

```sh

pip install rikai

make install

```

## Usage

To load `rikai` PostgreSQL extension

```sql

CREATE EXTENSION plpython3u;

CREATE EXTENSION rikai;

```

Create a model via `INSERT INTO`

```sql

INSERT INTO ml.models

(name, flavor, model_type, uri)

VALUES

('cat_detector', 'pytorch', 'ssd', 's3://bucket/to/cat.pth')

```

A function `ml.` will be created after model insertation.

To use the registered model for inference:

```sql

SELECT ml.cat_detector(image) FROM cat_dataset

```

Show all registered models:

```sql

SELECT * FROM ml.models;

```

## Local development

```sh

# Build and run docker image, prepare testing data and load a model.

$ make run

# Connect to postgres

$ psql -h localhost -U postgres

```

## TODOs

- [ ] Support classification models

- [ ] Release and installation

## Limitations

- `rikai` needs to be installed with the system python.

- Batch inference.

- Not ready for production yet.