https://github.com/justinchuby/onnxruntime-easy

Simplified APIs for onnxruntime
https://github.com/justinchuby/onnxruntime-easy

ai machine-learning onnx onnxruntime

Last synced: about 2 months ago
JSON representation

Simplified APIs for onnxruntime

• Host: GitHub
• URL: https://github.com/justinchuby/onnxruntime-easy
• Owner: justinchuby
• License: mit
• Created: 2025-03-18T23:34:53.000Z (3 months ago)
• Default Branch: main
• Last Pushed: 2025-03-19T00:39:57.000Z (3 months ago)
• Last Synced: 2025-03-19T01:25:48.296Z (3 months ago)
• Topics: ai, machine-learning, onnx, onnxruntime
• Language: Python
• Homepage:
• Size: 6.84 KB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# onnxruntime-easy

Simplified APIs for onnxruntime

## Usage

```py

import onnxruntime_easy as ort_easy

import numpy as np

import ml_dtypes

# Simple `load` method that handles setting up ONNX Runtime inference session

# All session options discoverable in the load function.

model = ort_easy.load("model.onnx", device="cpu")  # You can control the providers if the default is not what you need

# Supports all ONNX dtypes via ml_dtypes or dlpack

input = ort_easy.ort_value(np.random.rand(1, 3, 299, 299).astype(ml_dtypes.bfloat16))

output = model(input)

# Works with any ndarray that implements the __array__ interface

# Or automatically share data on device (like cuda) with dlpack

import torch

model = ort_easy.load("model.onnx", device="cuda")

input_tensor = ort_easy.ort_value(torch.rand(1, 3, 299, 299, device="cuda"))

output = model(input_tensor)

# Use a context manager to control the outputs you get

with model.set_outputs("output1"):

    output1 = model(input_tensor)

```