CLI utility to verify operator support before converting ONNX models to Kneron NEF format
onnx-check is a lightweight CLI that inspects ONNX models before you convert or deploy them. It tells you exactly which operators the graph uses and whether they are supported on your target hardware.
"Catch unsupported operators early โ before they derail your model." โ Mason Huang
| Feature | Description |
|---|---|
| Operator scan | Fast, dependencyโfree static analysis of .onnx files |
| Hardware profiles | Builtโin JSON compatibility tables for common NPUs (Currently only KL520, 530, 630, 720, 730 โฆ) with an easy override mechanism |
| Clear report | CLI table plus optional JSON / Markdown export; highlights unsupported ops and optionalโfeature gaps |
| Actionable hints | Suggestions and links to official docs for each unsupported operator |
| Milestone | Planned | Notes |
|---|---|---|
| 0.2 | Markdown/JSON report templates | Nice for CI bots |
| 0.3 | Model slimming (--prune, --quantize) | Reduce model size before flashing |
| 0.4 | Automatic op replacement (--replace) | Swap unsupported ops for functionallyโequivalent kernels |
| 0.5 | Interactive web viewer (onnx-op-view) | Dragโandโdrop visualiser |
| Command | Description |
|---|---|
pip install onnx-checker | Install latest package from PyPI |
onnxcheck my_model.onnx -p KL720 | Inspect my_model.onnx for the KL720 hardware profile |
onnxcheck my_model.onnx | Inspect my_model.onnx for all built-in profiles |
onnxcheck -V, onnxcheck --version | Show onnx-checker version |
Model summary โ my_model.onnx
IR version : 6 Opset : 11
Inputs : input float32 [1, 3, 112, 112]
Outputs : output float32 [1, 512]
Dynamic axes : None detected โ
my_model.onnx ยท IR 6 ยท KL520
โญโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฎ
โ Status Operator Count Notes โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโค
โ โ Conv 27 โ
โ โ Relu 27 โ
โ โ Elu 5 Not supported on KL520 โ
โ โ MaxPool 5 โ
โ โ Resize 2 Only linear/nearest modes OK โ
โฐโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฏ
โ 2 unsupported operator(s) detected.
from onnx_op_check import Checker, load_profile
checker = Checker("my_model.onnx", profile=load_profile("kl720"))
report = checker.run()
print(report.to_markdown())
Profiles live under onnx_op_check/profiles/*.json.
Each profile declares the operators, attributes, and constraints supported by a particular accelerator.
See docs/PROFILE_SCHEMA.md for the JSON schema.
Contributions for new hardware are very welcome!
We love pull requests! Please read CONTRIBUTING.md and open an issue before you start a large refactor so we can align on design.
Coding conventions follow PEPย 8 with the Black formatter.
Released under the MIT License ยฉย 2025ย Mason &ย contributors.
The primary language of this README is English for wider community reach. A Traditional Chinese translation will be added soon in docs/README_zh-TW.md.
FAQs
CLI utility to verify operator support before converting ONNX models to Kneron NEF format
We found that onnx-checker demonstrated a healthy version release cadence and project activity because the last version was released less than a year ago.ย It has 1 open source maintainer collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Research
Security News
Malicious npm packages targeting React, Vue, Vite, Node.js, and Quill remained undetected for two years while deploying destructive payloads.
Security News
Open source maintainers are urging GitHub to let them block Copilot from submitting AI-generated issues and pull requests to their repositories.
Research
Security News
Malicious Koishi plugin silently exfiltrates messages with hex strings to a hardcoded QQ account, exposing secrets in chatbots across platforms.