tencent cloud

Overview

TiKit is an open-source SDK for Python provided by TI Platform, which can be used to submit training tasks to TI Platform.

Directions

Environment dependencies

Currently, Python 3.4 or later is supported.

Installing TiKit

On TI Platform, TiKit is built in the containers for notebooks and training tasks, so you don't need to install it separately.

In a non-public cloud TI Platform environment, the installation methods are as follows:

• Install dependencies:

# CentOS:
sudo yum -y install cyrus-sasl cyrus-sasl-devel cyrus-sasl-lib
# Ubuntu:
sudo apt-get update
sudo apt-get install -y libsasl2-dev

• Method 1. Install through pip (recommended)

pip3 install -U tikit

• Method 2. Install offline. Download the installation package at https://pypi.org/project/tikit/ and install it with the .whl file or source code:

pip3 install tikit-1.0.0-py3-none-any.whl
# Alternatively, after decompressing the source code, run
python3 setup.py install

Getting started

1. Prepare the secret_id and secret_key.

   Log in to Tencent Cloud to get them on the corresponding page as shown below:

2. Initialize the task and call the function.

from tikit.client import Client
# Initialize the client. In a public cloud TI Platform environment (including notebooks and training tasks), you can leave the region information empty, which is already contained in the environment variable.
client = Client("your_secret_id", "your_secret_key", "<region>")
# View the list of algorithm frameworks.
client.describe_training_frameworks()

In a notebook environment, the HTML content will be directly displayed as shown below:

In a non-notebook environment, you can print the result as shown below:

3. View the function usage.

help(client.create_image_dataset)