A VS Code extension that automatically displays PyTorch tensor shapes and dtypes in the debugger's variable window.
Features
🔍 Automatic Shape Display: Shows tensor shapes like [2, 3, 4] directly in the debugger variable window
📊 Data Type Information: Displays dtype (e.g., torch.float32) alongside shapes
⚡ Zero Configuration: Works automatically when you start debugging Python code
🎯 Non-Intrusive: Only activates during Python debug sessions
How It Works
When you start debugging Python code, this extension automatically injects a small patch that modifies torch.Tensor.__repr__ to display shape and dtype information prominently in the debugger.
Before:
tensor([[1., 2., 3.]])
After:
[1, 3] torch.float32 | tensor([[1., 2., 3.]])
Usage
Install the extension
Start debugging your Python code that uses PyTorch
Check the Variables panel - tensor shapes will now be visible!
Manual Injection
If you need to inject the patch manually during a debug session:
Open the Command Palette (Cmd+Shift+P on macOS, Ctrl+Shift+P on Windows/Linux)
Run: Tensor Debugger Shape: Inject Patch
Configuration
You can configure the extension in VS Code settings:
tensorDebuggerShape.enabled: Enable/disable the extension (default: true)
tensorDebuggerShape.autoInject: Automatically inject patch on debug start (default: true)
Requirements
VS Code 1.70.0 or higher
Python with PyTorch installed
Python debugger extension (ms-python.python or ms-python.debugpy)
Development
To run the extension locally:
# Install dependencies
npm install
# Open in VS Code
code .
# Press F5 to open a new Extension Development Host window
Known Limitations
Only works with PyTorch tensors
Patch is applied per-session (not persistent across restarts)
May not work if torch is imported after debugging starts
License
MIT
Contributing
Contributions are welcome! Please feel free to submit issues or pull requests.
