pytorch/ignite

View on GitHub

High-level library to help with training and evaluating neural networks in PyTorch flexibly and transparently.

Stars: 4.8k
Forks: 726
Open beginner issues: 0
Indexed issues: 110
Avg merge: 5d 21h
Merged PRs (30d): 5

Dominant language: Python
License: BSD-3-Clause
Last GitHub push: Sep 9, 2026
Latest indexed: Sep 19, 2026
Contributing guide: Contributing guide
Code of conduct: Code of conduct
Beginner labels: help wanted hacktoberfest good first issue

Beginner friendly Everything

Open All

110 open issues indexed

Use monotonic time for training time limits Open

Difficulty 2/5 1-3 hours Newbie friendliness 72/100

pytorch/ignite#3841 · 1 comment · Opened Sep 12, 2026
Pass lists of scalars to wandb Open

Difficulty 4/5 3-5 days Newbie friendliness 48/100

pytorch/ignite#3809 · 2 comments · Opened Jul 15, 2026
Attaching every loss separately is different from attaching them with MetricGroup Open

Difficulty 3/5 1-2 days Newbie friendliness 52/100

pytorch/ignite#3806 · 6 comments · Opened Jul 10, 2026
Numerical stability across running-moment metrics — unified plan (tracking #3662 / #3672 / #3741) Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

pytorch/ignite#3748 · 5 comments · Opened May 11, 2026
FSDP Support Open

Difficulty 5/5 Over a week Newbie friendliness 38/100

pytorch/ignite#3724 · 8 comments · Opened Apr 6, 2026
Add GradientNormLogger handler for monitoring training stability Open

needs-discussion

Difficulty 3/5 1-2 days Newbie friendliness 55/100

pytorch/ignite#3707 · 8 comments · Opened Mar 30, 2026
Distributed may deadlock in `one_rank_first` when an exception occurs on the target rank Open

Difficulty 4/5 3-5 days Newbie friendliness 52/100

pytorch/ignite#3675 · 11 comments · Opened Mar 15, 2026
[Enhancement] Native PyTorch Implementation for `SpearmanRankCorrelation` Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

pytorch/ignite#3663 · 9 comments · Opened Mar 9, 2026
[Bug] Numerical instability in `PearsonCorrelation` due to naive variance formula Open

bug module: metrics

Difficulty 4/5 3-5 days Newbie friendliness 48/100

pytorch/ignite#3662 · 8 comments · Opened Mar 9, 2026
Feature Request: Option to disable automatic epoch display in ProgressBar Open

needs-discussion waiting-for-response

Difficulty 4/5 3-5 days Newbie friendliness 42/100

pytorch/ignite#3659 · 4 comments · Opened Mar 8, 2026
Feature Request: StratifiedBatchSampler for Guaranteed Subgroup Representation Open

Difficulty 5/5 Over a week Newbie friendliness 38/100

pytorch/ignite#3653 · 9 comments · Opened Mar 7, 2026
HarmonicMean metric for ignite.metrics Open

Difficulty 3/5 1-2 days Newbie friendliness 50/100

pytorch/ignite#3648 · 10 comments · Opened Mar 6, 2026
[Inconsistency]: Fbeta metric lacks is_multilabel and advanced average parameters Open

enhancement module: metrics

Difficulty 3/5 1-2 days Newbie friendliness 45/100

pytorch/ignite#3640 · 20 comments · Opened Mar 5, 2026
Bug: Events.GET_BATCH_STARTED(once=N) fires twice when engine.terminate_epoch(False) is called inside handler Open

module: engine

Difficulty 3/5 1-2 days Newbie friendliness 64/100

pytorch/ignite#3625 · 4 comments · Opened Mar 3, 2026
Resuming from checkpoint with max_epochs issue Open

module: engine needs-discussion question

Difficulty 3/5 1-2 days Newbie friendliness 48/100

pytorch/ignite#3622 · 7 comments · Opened Mar 2, 2026
[Feature]: Add Mean Reciprocal Rank (MRR) metric to rec_sys Open

Difficulty 3/5 1-2 days Newbie friendliness 45/100

pytorch/ignite#3569 · 1 comment · 2 reactions · Opened Feb 21, 2026
[Feature] Implement TopK wrapper class for metrics Open

Difficulty 3/5 1-2 days Newbie friendliness 35/100

pytorch/ignite#3568 · 13 comments · Opened Feb 21, 2026
[Feature] Implementation of a TrainingHealthMonitor for real-time gradient norm tracking. Open

Difficulty 5/5 Over a week Newbie friendliness 42/100

pytorch/ignite#3563 · 1 comment · Opened Feb 20, 2026
[Feature] Add reset_on_start to EarlyStopping Open

Difficulty 3/5 1-2 days Newbie friendliness 48/100

pytorch/ignite#3535 · 15 comments · Opened Feb 14, 2026
`idist.initialize` fails in Slurm when using `--ntasks-per-gpu` Open

Difficulty 4/5 3-5 days Newbie friendliness 57/100

pytorch/ignite#3259 · 8 comments · Opened Jun 26, 2024
INFO: Resuming from iteration for provided data will fetch data until required iteration ... Open

question

Difficulty 3/5 1-2 days Newbie friendliness 35/100

pytorch/ignite#3247 · 3 comments · Opened May 15, 2024
Calculation of FID metric made use of classification probabilities instead of feature vectors Open

question

Difficulty 2/5 1-3 hours Newbie friendliness 62/100

pytorch/ignite#3221 · 2 comments · Opened Mar 29, 2024
Event handler to "unlock" GCPs Vertex.AI hyperparameter tuning service Open

help wanted

pytorch/ignite#3023 · 3 comments · 1 assignee · Opened Aug 14, 2023
Workaround for torch.linalg lazy init for multi-GPU setup? Open

question

Difficulty 4/5 3-5 days Newbie friendliness 25/100

pytorch/ignite#3004 · 4 comments · 2 reactions · Opened Jul 23, 2023
Problem with using "mps" device Open

enhancement help wanted module: distributed

Difficulty 3/5 1-2 days Newbie friendliness 35/100

pytorch/ignite#2977 · 5 comments · Opened Jun 29, 2023
Addition of more tutorial notebooks on topics like Siamese Networks and Super Resolution techniques Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

pytorch/ignite#2878 · 8 comments · Opened Feb 19, 2023
Support for TorchSnapshot for efficient checkpoint saving and loading Open

enhancement help wanted

Difficulty 4/5 3-5 days Newbie friendliness 38/100

pytorch/ignite#2752 · 3 comments · Opened Oct 24, 2022
Sync trainer state with evaluators Open

enhancement Hacktoberfest help wanted needs-discussion

Difficulty 5/5 Over a week Newbie friendliness 30/100

pytorch/ignite#2733 · 10 comments · Opened Oct 6, 2022
Improvements for resuming from terminated engine Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

pytorch/ignite#2669 · 1 comment · Opened Aug 23, 2022
Upgrading to 0.4.9 stuck multi-gpu training Open

pytorch/ignite#2635 · 11 comments · 1 assignee · Opened Aug 4, 2022
Implementation of RecSys Metrics Open

enhancement help wanted

Difficulty 4/5 3-5 days Newbie friendliness 35/100

pytorch/ignite#2631 · 14 comments · Opened Jul 23, 2022
Checkpoint support for `torch.package` Open

Difficulty 4/5 3-5 days Newbie friendliness 38/100

pytorch/ignite#2570 · 3 comments · Opened May 13, 2022
How to resume learning? Open

question

Difficulty 3/5 1-2 days Newbie friendliness 25/100

pytorch/ignite#2569 · 10 comments · Opened May 12, 2022
how to use all_gather in training loop? Open

question

Difficulty 4/5 3-5 days Newbie friendliness 30/100

pytorch/ignite#2504 · 11 comments · Opened Mar 7, 2022
DataParallel is used by auto_model with single GPU Open

question

Difficulty 4/5 3-5 days Newbie friendliness 25/100

pytorch/ignite#2447 · 9 comments · 1 reaction · Opened Feb 2, 2022
Approximate PR/ROC AUC Metics Open

enhancement module: metrics

pytorch/ignite#2319 · 13 comments · 1 assignee · Opened Nov 11, 2021
Add no improvement handler (similar to early stopping handler) Open

enhancement help wanted

pytorch/ignite#2314 · 8 comments · 1 assignee · Opened Nov 9, 2021
Update pytorch.org/ignite, README, contributing docs Open

docs enhancement

pytorch/ignite#2302 · 3 comments · 1 assignee · Opened Oct 29, 2021
List of schedulers support in `setup_common_training_handlers` Open

Difficulty 3/5 1-2 days Newbie friendliness 35/100

pytorch/ignite#2289 · 4 comments · Opened Oct 22, 2021
Improve idist for gather using nccl and reduce for gloo + gpu Open

enhancement

Difficulty 4/5 3-5 days Newbie friendliness 35/100

pytorch/ignite#2260 · 1 comment · 1 reaction · Opened Oct 12, 2021
[FR] Add kwargs to `idist.barrier` Open

enhancement module: distributed

pytorch/ignite#2213 · 4 comments · 1 assignee · Opened Sep 24, 2021
🥇 Everyone is welcome to contribute 💯 Open

good first issue Hacktoberfest help wanted PyDataGlobal

Difficulty 5/5 Over a week Newbie friendliness 20/100

pytorch/ignite#2026 · 50 comments · 6 reactions · Opened Jun 2, 2021
MLFlow support in ignite Open

question

Difficulty 5/5 Over a week Newbie friendliness 25/100

pytorch/ignite#2011 · 2 comments · Opened May 24, 2021
Possible speed improvement Open

question

Difficulty 4/5 3-5 days Newbie friendliness 32/100

pytorch/ignite#2010 · 4 comments · 1 reaction · Opened May 22, 2021
DeepSpeed support for ignite.distributed Open

enhancement

Difficulty 5/5 Over a week Newbie friendliness 30/100

pytorch/ignite#2008 · 8 comments · 3 reactions · Opened May 20, 2021
Add `dict` properties to Engine and State Open

needs-discussion question

Difficulty 4/5 3-5 days Newbie friendliness 35/100

pytorch/ignite#2000 · 13 comments · 1 reaction · Opened May 17, 2021
Add device arg to Engine Open

enhancement needs-discussion

Difficulty 3/5 1-2 days Newbie friendliness 45/100

pytorch/ignite#1999 · 3 comments · 1 reaction · Opened May 16, 2021
Debug mode for ignite Open

enhancement help wanted

Difficulty 5/5 Over a week Newbie friendliness 20/100

pytorch/ignite#1992 · 15 comments · 2 reactions · Opened May 7, 2021
ModelCheckpoint improvement from "remove & write" to "write & remove" Open

enhancement question

Difficulty 5/5 Over a week Newbie friendliness 25/100

pytorch/ignite#1986 · 14 comments · Opened May 4, 2021
Agnostic Engine configured by specific handlers Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

pytorch/ignite#1949 · Opened Apr 19, 2021

pytorch/ignite

Get new issues in your inbox