Python API

Managing Tensor One Resources via SDK.

This document outlines the core functionalities provided by the Tensor One Python SDK, including how to interact with endpoints, manage templates, and list available GPUs. These operations allow you to dynamically manage computational resources within the Tensor One environment.

Get Endpoints

Retrieve a list of all available endpoint configurations.

import tensorone
import os

tensorone.api_key = os.getenv("TENSORONE_API_KEY")

# Fetching all available endpoints
endpoints = tensorone.get_endpoints()
print(endpoints)

Create a Template

Create a new template by specifying a name and Docker image.

import tensorone
import os

tensorone.api_key = os.getenv("TENSORONE_API_KEY")

try:
    new_template = tensorone.create_template(
        name="test",
        image_name="tensorone/base:0.1.0"
    )
    print(new_template)

except tensorone.error.QueryError as err:
    print(err)
    print(err.query)

Create an Endpoint

Create a new endpoint with a specific template and GPU configuration.

import tensorone
import os

tensorone.api_key = os.getenv("TENSORONE_API_KEY")

try:
    new_template = tensorone.create_template(
        name="test",
        image_name="tensorone/base:0.4.4",
        is_serverless=True
    )
    print(new_template)

    new_endpoint = tensorone.create_endpoint(
        name="test",
        template_id=new_template["id"],
        gpu_ids="AMPERE_16",
        workers_min=0,
        workers_max=1,
    )
    print(new_endpoint)

except tensorone.error.QueryError as err:
    print(err)
    print(err.query)

Get Available GPUs

List all available GPUs for endpoint allocation.

import tensorone
import json
import os

tensorone.api_key = os.getenv("TENSORONE_API_KEY")

gpus = tensorone.get_gpus()
print(json.dumps(gpus, indent=2))

Get GPU by ID

Retrieve details of a specific GPU using its ID.

import tensorone
import json
import os

tensorone.api_key = os.getenv("TENSORONE_API_KEY")

gpu = tensorone.get_gpu("NVIDIA A100 80GB PCIe")
print(json.dumps(gpu, indent=2))

Through these functionalities, the Tensor One Python SDK enables efficient and flexible management of compute resources to support a wide range of project needs.