The MLOps platform for innovators 🚀

Last update: Jan 03, 2023

Overview

The official DS2.ai SDK for Python.
Documentation can be found on SDK guide

MLOps with DS2.ai

DS2.ai is an integrated AI operation solution that supports all stages from custom AI development to deployment. It is an AI-specialized platform service that collects data, builds a training dataset through data labeling, and enables automatic development of artificial intelligence and easy deployment and operation. The Software Development Kit (SDK) consists of python functions that allow you to write your own scripts by accessing DS2.ai's features.

Installation

Install via pip:

$ pip install ds2ai

Getting started

1. Getting your own token

To use the SDK, you need to get a token, and you can check the token by registering as a member of ds2.ai. After registering the card on the site, you can use the token.

### 2. Activate To use SDK function code, you have to activate your code, first. Run the below code with your own app token. ```python import ds2ai ds2 = ds2ai.DS2(token) ``` Then you can use all functions in [SDK guide](https://docs.ds2.ai/sdk_00_readme/) --- ## Top 5 Features of [DS2.ai](https://ds2.ai/) SDK The SDK is composed of 16 classes. Class DS2 provides python functions that are more generally used for AI development, whereas the others provide specific functions for each detailed steps in AI development. Here, we want to explain to you examples of using **Top5 function codes that are usable and easy to use.** ### 1. Auto Labeling ```python ds2.start_auto_labeling(data_file, amount, has_label_data=False, predict_column_name=None, frame=60, ai_type="general", autolabeling_type="box", general_ai_type="person", model_id=None, custom_ai_stage=0, preprocessing_ai_type={}, labeling_class=[], training_method="object_detection", name='', description='' )

### 2. AI Training ```python ds2.train(data_file, training_method, value_for_predict, option="accuracy", frame=60) ">


This function executes auto-labeling immediately from loading data file without using dataconnector. The major parameters include data_file to auto-label, whether the data includes labeled data for a certain part of the dataset, and the type of auto-labeling, such as “box”, which will label using bounding boxes.





### 2. AI Training

```python
ds2.train(data_file, training_method, value_for_predict, option="accuracy", frame=60)

This function executes development of AI from CLICK AI in DS2.ai’s console immediately from loading data file without using dataconnector. According to what parameters you use when calling the function, such as data_file, training_method, value_for_predict, and option, it will generate your customized AI models.

3. Deploy your AI model

ds2.deploy(model_file, name=None, cloud_type="AWS", region="us-west-1", server_type="g4dn.xlarge")

This function deploys AI models to cloud servers with specifications under the desired hosting region. The type of the cloud server is set to “AWS” as default, but keep in mind that it also supports other cloud services such as Google Cloud. For the use of servers other than AWS, please visit our website and contact our team.

### 4. Getting magic code ```python ds2.get_magic_code(training_method, data_file, value_for_predict) ```

This function returns a the magic code for setting variable values with optimal combinations for AI training. As with the three functions above, it takes the data_file, training_method, value_for_predict as input so that after running the function, a magic code with the whole process of AI training is returned.

5. Rent AI training server

ds2.rent_custom_training_server(cloud_type="AWS", region="us-west-1", server_type="g4dn.xlarge", name=None)

This function rents an inference training server in preferred cloud environment for Custom training of Click AI. The type of the cloud server is set to “AWS” as default, but keep in mind that it also supports other cloud services such as Google Cloud. For the use of servers other than AWS, please visit our website and contact our team.

Getting Help

You can interact with the ds2ai code or software by asking a question or referencing the guide from the underlying open resources.

GitHub
- https://github.com/DS2BRAIN/ds2ai
stackoverflow - Ask all code and software related questions here.
- https://stackoverflow.com/questions/tagged/ds2ai
CrossValidated - Ask questions about artificial intelligence algorithms and theories here.
- https://stats.stackexchange.com/questions/tagged/ds2ai
Documentation
- ds2ai user guide : https://docs.ds2.ai/
- ds2ai blog : https://blog.ds2.ai/
Website
- https://ds2.ai/ If you need help that is not specific to this SDK, please reach out to the chat "Ask us in" in our application.

License

This SDK is distributed under the Apache-2.0 License, please see LICENSE for more information.

The MLOps platform for innovators 🚀

Related tags

Overview

MLOps with DS2.ai

Installation

Getting started

1. Getting your own token

3. Deploy your AI model

5. Rent AI training server

Getting Help

License

Owner

RCT-ART is an NLP pipeline built with spaCy for converting clinical trial result sentences into tables through jointly extracting intervention, outcome and outcome measure entities and their relations.

Usable Implementation of "Bootstrap Your Own Latent" self-supervised learning, from Deepmind, in Pytorch

The code is the training example of AAAI2022 Security AI Challenger Program Phase 8: Data Centric Robot Learning on ML models.

XViT - Space-time Mixing Attention for Video Transformer

ECAENet (TensorFlow and Keras)

Implementation based on Paper - Learning a Probabilistic Latent Space of Object Shapes via 3D Generative-Adversarial Modeling

GemNet model in PyTorch, as proposed in "GemNet: Universal Directional Graph Neural Networks for Molecules" (NeurIPS 2021)

PyTorch version of the paper 'Enhanced Deep Residual Networks for Single Image Super-Resolution' (CVPRW 2017)

A dead simple python wrapper for darknet that works with OpenCV 4.1, CUDA 10.1

Python with OpenCV - MediaPip Framework Hand Detection

The codes of paper 'Active-LATHE: An Active Learning Algorithm for Boosting the Error exponent for Learning Homogeneous Ising Trees'

An image classification app boilerplate to serve your deep learning models asap!

Image-Scaling Attacks and Defenses

The Implicit Bias of Gradient Descent on Generalized Gated Linear Networks

The source code for Generating Training Data with Language Models: Towards Zero-Shot Language Understanding.

Patch SVDD for Image anomaly detection

FlowTorch is a PyTorch library for learning and sampling from complex probability distributions using a class of methods called Normalizing Flows

YOLOv5 🚀 is a family of object detection architectures and models pretrained on the COCO dataset

This repository is based on Ultralytics/yolov5, with adjustments to enable polygon prediction boxes.

Code for Subgraph Federated Learning with Missing Neighbor Generation (NeurIPS 2021)