# maxpumperla/elephas

Distributed Deep learning with Keras & Spark

Repository: https://github.com/maxpumperla/elephas
Canonical: https://ross.abutalabs.com/products/elephas
Homepage: http://maxpumperla.com/elephas/
Language: Python
License: MIT
License Family: permissive
Topics: spark, keras, neural-networks, deep-learning, distributed-computing
Last push: 2023-05-01T20:46:16+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 4038, "days_push": 1220, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1577, forks 303 (observed 2026-08-28T04:05:06.167295+00:00)

## What it is
Elephas is a Python library that extends Keras to run distributed deep learning training on Apache Spark. It serializes Keras models on the Spark driver, ships them to workers for data-parallel training, and aggregates gradients back on the driver.

## Use cases
- train keras models on spark clusters
- distributed deep learning on large datasets
- data-parallel training of neural networks with spark rdds
- integrate keras models into spark ml pipelines
- distributed training of ensemble models
- use keras with spark mllib

## When to choose
- you already use keras and need to scale training across a spark cluster
- your training data lives in spark rdds or dataframes
- you want keras-style simplicity for distributed model prototyping

## When to avoid
- you need actively maintained software - development moved to a fork (danielenricocahall/elephas)
- you need distributed hyper-parameter optimization, which was removed in 3.0.0
- you use pytorch or tensorflow's native distributed strategies instead of keras on spark

## Facets
- artifact type: library
- maturity: maintenance
- function: deep-learning, machine-learning, etl
- domain: deep-learning, machine-learning, big-data, microservices
- platform: python, jvm
- tags: keras, spark, distributed-training, data-parallel, rdd, mllib

## Member repositories
- maxpumperla/elephas (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:06.167295+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 2026-08-30T03:57:09.944720+00:00, confidence not recorded.
  - readme: https://github.com/maxpumperla/elephas (fetched 2026-08-28T04:05:06.167295+00:00, sha 0b0ce16d54d5)
- Data as of 2026-08-30T08:39:29.467469+00:00.