fancy-einsum

library0.0.3pypypi✓ verified 85d ago

A drop-in replacement for torch/numpy einsum that allows descriptive variable names in equations. Current version 0.0.3. Infrequently updated; released in 2021.

ai-ml

pip install fancy-einsum

github ↗homepage ↗

INSTALL

IMPORT

SIG · FANCY-EINSUM

fancy-einsum

ai-mlpythonv0.0.3

harness data pending

Install & Compatibility

Where this runs

No compatibility data collected yet for this library.

Code

Verified usage

Verified import paths — ran on the pinned version, not inferred.

einsum

✓ from fancy_einsum import einsum

✗ from fancy_einsum import fancy_einsum

get_backend

✓ from fancy_einsum import get_backend

convert_equation

✓ from fancy_einsum import convert_equation

Einstein summation with descriptive axis names instead of letters.

import torch from fancy_einsum import fancy_einsum x = torch.randn(3, 4) y = torch.randn(4, 5) # Using descriptive names: 'batch seq' -> 'batch hidden' z = fancy_einsum('batch seq, seq hidden -> batch hidden', x, y) print(z.shape) # torch.Size([3, 5])

Debug

Known issues

gotchaVariable names must be separated by spaces; repeated names imply contraction.

fix

Ensure spaces between names and that contraction axes use the same name.

affects: all

gotchaDoes not support ellipsis (...) for batch dimensions; use explicit names.

fix

Replace '...' with explicit dimension names like 'batch'.

affects: all

deprecatedLibrary is not actively maintained; last release May 2021. May not support newer PyTorch versions.

fix

Consider using einops.einsum with named axes or torch.einsum with named tensors.

affects: >=0.0.3

Upgrade

Version history

0.0.3latest on PyPI · released Feb 4, 2022

Audit

Dependencies

No dependency data recorded yet.

Agent activity

17 hits · last 30 days

node

OpenAI (training)

Resources

githubhttps://github.com/Kiv/fancy_einsum ↗

homepagehttps://github.com/Kiv/fancy_einsum ↗