NPZ Viewer
A lightweight VS Code extension for viewing NumPy .npy and .npz files - displays variable names, shapes, and data types with zero dependencies beyond a single ZIP library.
Features
- Instant Preview - Open any
.npy or .npz file and see its contents immediately
- Dual Format Support - Handles both single array files (
.npy) and multi-file archives (.npz)
- Detailed Info - Shows variable name, shape, and dtype for each array
- Human-Readable Types - Dtypes are translated (e.g.,
<i4> → <i4> (int32, little-endian))
- Endianness Display - Shows little-endian, big-endian, or native byte order
- Data Preview - Click variable name to expand/collapse array data
- Statistics - Min, max, mean for numeric arrays (arrays up to 1M elements)
- Standard Deviation - For arrays ≤ 100 elements
- First 100 Values - Preview of array data as flat list or table
- Minimal Dependencies - Only requires
adm-zip (~50KB)
Installation
- Install from VS Code Marketplace (search "NPZ Viewer")
- Or install manually: download
.vsix file and run "Extensions: Install from VSIX..."
Usage
- Open any
.npy or .npz file in VS Code
- The extension automatically displays a table with:
- Variable - Name of array (or "array" for single
.npy files)
- Shape - Array dimensions (e.g.,
[100, 28, 28] or [scalar] for 0-dimensional arrays)
- Dtype - Data type with human-readable explanation
- Click any variable name to expand/collapse:
- Statistics (min, max, mean, std dev) for numeric arrays
- First 100 array values (formatted for readability)
Supported Dtypes
Supports all NumPy dtypes:
- Integers:
int8, int16, int32, int64
- Unsigned:
uint8, uint16, uint32, uint64
- Floats:
float16, float32, float64
- Complex, bytes, strings, unicode, void, objects
Shows endianness (little-endian, big-endian, native).
Requirements
- VS Code 1.96.0 or higher
- No Python or NumPy installation required
License
MIT License - see LICENSE for details.
| |
