tensorViz

tensorViz

Let your coding agent see inside your PyTorch model

A free tool that turns a PyTorch model into an interactive graph your coding agent can search and reason about.

tensorViz on Product Hunt

tensorViz on Product Hunt

1 / 2

Overview

What is tensorViz?

Let your coding agent see inside your PyTorch model

tensorViz lets your coding agent see inside your PyTorch model. Coding agents can read PyTorch code, but they cannot see the data actually flowing through the model. tensorViz runs one real forward pass on a local nn.Module and produces an interactive graph of every module and tensor shape. Claude Code, Codex, and Cursor can search the graph, read measured shapes, and fix model code from evidence instead of guessing. The Model Explorer lets you browse 560+ model graphs from recent papers without installing anything. Graphs contain only structure and shapes, never code or weights, so they stay private or can be shared by link. Free for everyone, with no account, email, tracking, or usage cap, and open source under the MIT license.

Platforms and languages

tensorViz Availability

Platforms

WebMacosWindowsLinux

Languages

English

Capabilities

tensorViz Key Features

Interactive model graph

One real forward pass produces a graph of every module and tensor shape.

Agent-searchable

Coding agents search layers, operators, and parameters and read measured shapes.

Drill down layer by layer

Zoom from the whole model to a single layer, on light or dark canvas.

Model Explorer

Browse 560+ model graphs from recent papers, filterable by field, size, and date.

Private by design

Graphs hold structure and shapes only, never code or weights.

Best for

Who uses tensorViz?

Machine learning engineers debugging PyTorch models with coding agents

Give the agent a measured model graph so it fixes code from evidence, not guesses

Plans and access

tensorViz Pricing

Free

No free trial listed

Common questions

tensorViz FAQs

How does tensorViz build the graph?

It runs one real forward pass on your local model and turns the result into an interactive graph.

How do I install it?

Install with pip install tensorviz. There are also editor and agent plugins, plus an MCP server.

Does my model leave my machine?

No. Models run locally, and graphs contain only structure and shapes, not code or weights.

Is it really free?

Yes. All features are free, with no account, API key, or usage cap.

Which coding agents does it support?

Claude Code, Codex, Cursor, and more, plus a VS Code extension and an MCP server.

Reviews

See what the community thinks and share your experience.

—

Based on 0 ratings

Rating distribution

0
0
0
0
0

Leave a review

Sign in to rate

Community reviews

No reviews

No written reviews yet.

Explore by category

AI Coding Assistants

View all AI Coding Assistants websites

Explore by category

AI Development Infrastructure

View all AI Development Infrastructure websites