Enterprise DNA
O Open Source Observability medium

Zetane Viewer

by Community

ML models and internal tensors 3D visualizer

ZV

OSS

Zetane Viewer

Added 1 June 2026

Overview

Zetane Viewer is an open-source Python library that visualizes machine learning models and their internal tensor operations in 3D space. It renders neural network architectures as interactive 3D graphs, allowing users to inspect layer shapes, activations, and data flow. The tool is community-maintained and available on GitHub.

Best for

Best for
Researchers and developers who need to visually inspect and debug deep learning model internals

Use cases

  • Debugging model internals by inspecting tensor shapes and activations
  • Understanding complex neural network architectures through 3D visualization
  • Presenting or teaching model structure in an intuitive 3D format

Notes

Zetane Viewer is an open-source Python library that visualizes machine learning models and their internal tensor operations in 3D space. It renders neural network architectures as interactive 3D graphs, allowing users to inspect layer shapes, activations, and data flow. The tool is community-maintained and available on GitHub.

1,804 stars on GitHub. Last updated 2022-08-08.

Use cases

  • Debugging model internals by inspecting tensor shapes and activations
  • Understanding complex neural network architectures through 3D visualization
  • Presenting or teaching model structure in an intuitive 3D format

Pros

  • Free and open-source with an active community
  • 3D visualization provides an intuitive grasp of model depth and data flow
  • Works directly with Python and common deep learning frameworks

Cons

  • Limited to Python ecosystem; no support for other languages
  • 3D rendering can be overwhelming for simple or linear models
  • May not support all custom layer types or exotic architectures

Indexed from awesome-llmops and enriched against its public facts.

Pros

  • Free and open-source with an active community
  • 3D visualization provides an intuitive grasp of model depth and data flow
  • Works directly with Python and common deep learning frameworks

Cons

  • Limited to Python ecosystem; no support for other languages
  • 3D rendering can be overwhelming for simple or linear models
  • May not support all custom layer types or exotic architectures