# tensorflow/fold

Deep learning with dynamic computation graphs in TensorFlow

Repository: https://github.com/tensorflow/fold
Canonical: https://ross.abutalabs.com/products/fold
Language: Python
License: Apache-2.0
License Family: permissive
Archived: true
Last push: 2021-06-26T15:49:57+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3514, "days_push": 1894, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, archived
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1815, forks 262 (observed 2026-08-28T04:05:39.877928+00:00)

## What it is
TensorFlow Fold is a Python library for building TensorFlow models that consume structured data, where the computation graph structure depends on the input. It implements dynamic batching, transforming batches of arbitrarily shaped graphs into a single static graph for efficient execution.

## Use cases
- build treelstm models for sentiment analysis on parse trees
- train recursive neural networks over arbitrary tree structures
- run deep learning models on variable-sized structured inputs
- batch differently shaped computation graphs efficiently in tensorflow
- process hierarchical data like parse trees with neural networks

## When to choose
- you need dynamic computation graphs over tree-structured data in tensorflow 1.x
- you want to implement recursive neural networks with efficient batching

## When to avoid
- you use modern tensorflow 2.x or pytorch, which support dynamic graphs natively
- you need an actively maintained library - the last release was 2021 and the project is inactive

## Facets
- artifact type: library
- maturity: abandoned
- function: deep-learning, machine-learning
- domain: deep-learning, machine-learning
- platform: python
- tags: tensorflow, dynamic-computation-graphs, dynamic-batching, recursive-neural-networks, treelstm, structured-data, natural-language-processing

## Member repositories
- tensorflow/fold (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:39.877928+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:20:39.494926+00:00, confidence not recorded.
  - readme: https://github.com/tensorflow/fold (fetched 2026-08-28T04:05:39.877928+00:00, sha 65017b0d4797)
- Data as of 2026-08-30T08:39:29.467469+00:00.