tensi 0.3.0

Interactive tensor shape visualization library

Tensi

Interactive tensor shape visualization

Tensi turns multi‑dimensional tensors into intuitive, interactive geometry so you can instantly see what your data looks like.

I built Tensi while wrangling distributed sharding‑and‑concatenation of latent tensors. Interactive geometric visuals of tensor shapes finally made the process make sense.

Installation

pip install tensi

Quick Start

import tensi
import torch

# Visualize a 3D tensor
tensor = torch.randn(2, 4, 5)  # [Batch, Rows, Cols]
fig = tensi.visualize(tensor)
fig.show()

Usage Examples

Basic Usage

import tensi
import torch
import numpy as np

# Using the quick visualize function
tensor = torch.randn(3, 4, 5)
fig = tensi.visualize(tensor, title="My Tensor")
fig.show()

# Using different data types
data = [[1, 2, 3], [4, 5, 6]]
fig = tensi.visualize(data, dtype="int32")
fig.show()

# Using numpy arrays
np_array = np.random.rand(2, 3, 4)
fig = tensi.visualize(np_array, dtype="float64")
fig.show()

Tensor Shapes

# 2D Tensor
tensor_2d = torch.randn(5, 7)  # [Rows, Cols]
fig = tensi.visualize(tensor_2d, title="2D Tensor")
fig.show()

# 3D Tensor
tensor_3d = torch.randn(3, 5, 7)  # [Batch, Rows, Cols]
fig = tensi.visualize(tensor_3d, title="3D Tensor - 3 Batches")
fig.show()

# 4D Tensor
tensor_4d = torch.randn(2, 4, 5, 6)  # [Batch, Width, Height, Depth]
fig = tensi.visualize(tensor_4d, title="4D Tensor - 2 Batches")
fig.show()

Save Visualizations

# Save as interactive HTML
fig = tensi.visualize(tensor)
fig.write_html("tensor_visualization.html")

# Save as static image (requires kaleido)
fig.write_image("tensor_visualization.png")

Requirements

• Python >= 3.7
• PyTorch >= 1.8.0
• NumPy >= 1.19.0
• Plotly >= 5.0.0

Contributing

Contributions are welcome! Please feel free to submit a PR.

Links

• Github
• PyPI