#4444th of 68 in Training and fine-tuning

polyaxon

Kubernetes platform for tracking, running and tuning machine learning experiments

Stars
3.7k
License
Apache-2.0
Last commit
Nov 2026
Language
MDX
Stars, last 29 weeks+58
25 Apr 2026: 3,690 stars2 May 2026: 3,693 stars9 May 2026: 3,694 stars16 May 2026: 3,695 stars23 May 2026: 3,696 stars30 May 2026: 3,697 stars6 Jun 2026: 3,699 stars13 Jun 2026: 3,699 stars20 Jun 2026: 3,700 stars27 Jun 2026: 3,703 stars4 Jul 2026: 3,706 stars11 Jul 2026: 3,709 stars18 Jul 2026: 3,710 stars25 Jul 2026: 3,711 stars1 Aug 2026: 3,712 stars8 Aug 2026: 3,713 stars15 Aug 2026: 3,713 stars22 Aug 2026: 3,716 stars29 Aug 2026: 3,721 stars5 Sept 2026: 3,723 stars12 Sept 2026: 3,724 stars19 Sept 2026: 3,727 stars26 Sept 2026: 3,731 stars3 Oct 2026: 3,733 stars10 Oct 2026: 3,738 stars17 Oct 2026: 3,742 stars24 Oct 2026: 3,744 stars31 Oct 2026: 3,745 stars7 Nov 2026: 3,747 stars14 Nov 2026: 3,748 stars

3,690 on 25 Apr 20263,748 on 14 Nov 2026

From GitHub's daily star counts.

Checked on GitHub.

Overview

Polyaxon is a Kubernetes-based platform that runs and tracks ML training jobs, notebooks and TensorBoard from a CLI and a web dashboard. It includes hyperparameter search (grid, random, Hyperband, Bayesian, Hyperopt), distributed job operators for TensorFlow, PyTorch, MPI, Spark and Dask, and container-native DAG pipelines. It targets teams sharing GPU servers.

Who it is for: ML teams sharing GPU servers on Kubernetes for experiment tracking and training

Strengths

  • Built-in hyperparameter search: grid, random, Hyperband, Bayesian optimization, Hyperopt and custom iterative
  • Distributed training via operators for TensorFlow, PyTorch, MPI, Horovod, Spark and Dask
  • DAG engine runs jobs, services, distributed jobs and nested DAGs as pipeline steps
  • CLI and dashboard launch Jupyter notebooks and TensorBoard per project

Weaknesses

  • Deployment needs a Kubernetes cluster and Helm; no single-machine path described
  • README lists TensorFlow, MXNet, Caffe and Torch; no LLM-specific tooling mentioned
  • Distributed jobs need a per-framework operator and code changes before they work
  • No GitHub release published, though the README badge shows v2.16.1

What it needs

  • GPU optional
  • Needs Kubernetes, Helm, kubectl
  • Models: TensorFlow, PyTorch, MXNet, Caffe, Torch

Also in Training and fine-tuning

See all 68
Also in Training and fine-tuning
RankProjectScore
1peftPython library for LoRA and other parameter-efficient fine-tuning of Transformers models Live demo ↗ (opens in a new tab)21.9k starsApache-2.089 out of 100
2unslothDesktop and web app to run and fine-tune local models Live demo ↗ (opens in a new tab)79.7k starsApache-2.088 out of 100
3trlPost-training library with SFT, DPO, GRPO and KTO trainers19.8k starsApache-2.088 out of 100
#4deepspeedPyTorch library for distributed training of very large models with ZeRO43.3k starsApache-2.087 out of 100
#5megatron-lmDistributed training library for large transformer models on NVIDIA GPUs18.4k starscustom license86 out of 100