#3636th of 60 in Training and fine-tuning
ignite
PyTorch library for training and evaluating models with engines, events and metrics
Live demo ↗
(opens in a new tab)Documentation ↗
(opens in a new tab)Website ↗
(opens in a new tab)Repository on GitHub ↗
(opens in a new tab)
- Stars
- 4.8k
- License
- BSD-3-Clause
- Last commit
- Nov 2026
- Last release
- Jul 2026
- Language
- Python
4,729 on 22 Apr 20264,802 on 11 Nov 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Ignite is a Python library on top of PyTorch that replaces hand-written epoch and iteration loops with an Engine object plus event handlers. It ships metrics (Precision, Recall, Accuracy, Confusion Matrix, IoU, about 20 regression metrics) and handlers for saving artifacts and logging. It is used as a library, so the training script keeps control flow.
Who it is for: PyTorch users who want less training-loop boilerplate with full control
Strengths
- Handlers can be any callable, with no base class to inherit
- Event filtering (every, once, custom filter), stacking with |, and custom events
- Metrics can be composed with arithmetic and torch methods
- Library approach: no inversion of control over your training script
Weaknesses
- Tied to PyTorch; no support for other frameworks
- Full-featured template examples are listed as coming soon
- Reproducible example logs on the Ignite Trains server are listed as coming soon
- You still write the train step yourself; it is not a turnkey trainer
What it needs
- GPU optional
- Docker
- Needs PyTorch
- Models: PyTorch models
What changed
From the release notes. No flag means the note does not say so, not that the upgrade is safe.
Also in Training and fine-tuning
See all 60| Rank | Project | Score |
|---|---|---|
| 1 | peftPython library for LoRA and other parameter-efficient fine-tuning of Transformers models | 89 out of 100 |
| 2 | unslothDesktop and web app to run and fine-tune local models | 88 out of 100 |
| 3 | trlPost-training library with SFT, DPO, GRPO and KTO trainers | 87 out of 100 |
| #4 | deepspeedPyTorch library for distributed training of very large models with ZeRO | 86 out of 100 |
| #5 | megatron-lmDistributed training library for large transformer models on NVIDIA GPUs | 85 out of 100 |