# higgsfield-ai/higgsfield

Fault-tolerant, highly scalable GPU orchestration, and a machine learning framework designed for training models with billions to trillions of parameters

Repository: https://github.com/higgsfield-ai/higgsfield
Canonical: https://ross.abutalabs.com/products/higgsfield
Language: Jupyter Notebook
License: Apache-2.0
License Family: permissive
Topics: cluster-management, deep-learning, distributed, llama, llama2, llm, machine-learning, mlops, pytorch
Last push: 2024-05-25T17:43:07+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": 3021, "days_push": 830, "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 4106, forks 705 (observed 2026-08-28T04:08:35.607016+00:00)

## What it is
Higgsfield is an open-source GPU orchestration and machine learning framework for fault-tolerant, distributed training of very large models (billions to trillions of parameters), such as LLMs. It manages node allocation, experiment queuing, and deployment via GitHub and GitHub Actions while exposing a standard PyTorch/DeepSpeed workflow.

## Use cases
- train a 70B LLaMA model across multiple GPUs
- orchestrate GPU clusters for deep learning experiments
- run distributed training with ZeRO-3 sharding
- queue and manage large-scale training experiments
- automate ML training pipelines with GitHub Actions
- fine-tune LLMs on a multi-node cluster

## When to choose
- you need to train or fine-tune very large language models across multiple nodes
- you want fault-tolerant GPU orchestration with experiment queuing
- you prefer a plain PyTorch workflow with DeepSpeed/FSDP sharding
- you want training runs triggered and tracked through GitHub

## When to avoid
- you only train small models on a single GPU
- you need a fully managed commercial training platform with support
- you need Kubernetes-native scheduling rather than GitHub-based deployment
- you require a project with frequent recent releases

## Facets
- artifact type: framework
- maturity: maintenance
- function: llm-training, gpu-computing, machine-learning, deep-learning, workflow-automation, scheduling
- domain: deep-learning, large-language-models, machine-learning, gpu-computing
- platform: python, cloud
- tags: distributed-training, deepspeed, zero-3, pytorch, mlops, cluster-management, llama, github-actions, devops, docker, gpu, linux

## Member repositories
- higgsfield-ai/higgsfield (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:35.607016+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 2026-08-29T18:23:13.769073+00:00, confidence not recorded.
  - readme: https://github.com/higgsfield-ai/higgsfield (fetched 2026-08-28T04:08:35.607016+00:00, sha 2fe26ba947ca)
- Data as of 2026-08-30T08:39:29.467469+00:00.