# twitter-research/tgn

TGN: Temporal Graph Networks

Repository: https://github.com/twitter-research/tgn
Canonical: https://ross.abutalabs.com/products/tgn
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2024-05-28T16:05:00+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2232, "days_push": 827, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1202, forks 246 (observed 2026-08-28T04:03:58.517223+00:00)

## What it is
Official PyTorch implementation of Temporal Graph Networks (TGN), a framework for deep learning on dynamic graphs represented as sequences of timed events. It includes training scripts for self-supervised link prediction and supervised node classification, plus baseline comparisons on datasets like Wikipedia and Reddit.

## Use cases
- implement temporal graph networks for dynamic graph learning
- run link prediction on timestamped interaction datasets
- train models on dynamic node classification tasks
- reproduce TGN paper experiments and baselines
- compare memory-based models on dynamic graphs
- benchmark deep learning models on JODIE datasets

## When to choose
- you need state-of-the-art models for learning on dynamic or temporal graphs
- you want to reproduce or extend the TGN paper's results
- your data is a sequence of timed events or interactions between nodes
- you need a research baseline for temporal link prediction

## When to avoid
- you need a production-ready, well-maintained library with long-term support
- you work with static graphs rather than time-stamped event sequences
- you need a high-level API or integration with modern graph frameworks like PyTorch Geometric or DGL
- you require recent PyTorch versions, since the code targets older dependencies

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning
- domain: machine-learning, data-science
- platform: python
- tags: temporal-graphs, dynamic-graphs, graph-neural-networks, research-code, pytorch, link-prediction, algorithms

## Member repositories
- twitter-research/tgn (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:58.517223+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 2026-08-30T06:20:38.628331+00:00, confidence not recorded.
  - readme: https://github.com/twitter-research/tgn (fetched 2026-08-28T04:03:58.517223+00:00, sha 0f75813e5b88)
- Data as of 2026-08-30T08:39:29.467469+00:00.