Benchmark Suite Results
MLPerf Training
The MLPerf Training benchmark suite measures how fast systems can train models to a target quality metric. Current and previous results can be reviewed through the results dashboard below.
The MLPerf Training benchmark paper provides a detailed description of the motivation and guiding principles behind the MLPerf Training benchmark suite.
Results
MLCommons results are shown in an interactive table to enable you to explore the results. You can apply filters to see just the information you want and click across the top tabs to view the results visually. To see all result details, expand the columns by clicking on the “+” icon, which appears when you hover over “System Name” and subsequent columns.
Benchmarks
Each benchmark is defined by a Dataset and Quality Target. The following table summarizes the benchmarks in this version of the suite. The rules remain the official source of truth.
| Area | Benchmark | Dataset | Quality Target | Reference Implementation Model | Latest Version Available |
|---|---|---|---|---|---|
| Vision | Image classification | ImageNet | 75.90% classification | ResNet-50 v1.5 | v4.1 |
| Vision | Image segmentation (medical) | KiTS19 | 0.908 Mean DICE score | 3D U-Net | v4.1 |
| Vision | Object detection (light weight) | Open Images | 34.0% mAP | RetinaNet | v4.1 |
| Language | NLP | Wikipedia 2020/01/01 | 0.72 Mask-LM accuracy | BERT-large | v4.1 |
| Language | LLM | C4 | 2.69 log perplexity | GPT3 | v4.1 |
| Language | LLM finetuning | SCROLLS GovReport | 0.925 cross entropy loss | Llama 2 70B | v4.1 |
| Commerce | Recommendation | Criteo 4TB multi-hot | 0.8032 AUC | DLRM-dcnv2 | v4.1 |
| Marketing, Art, Gaming | Image Generation | LAION-400M-filtered | FID<=90 and CLIP>=0.15 | Stable Diffusionv2 | v4.1 |
| Graph neural network | Graph neural network (GNN)* | IGBH-Full | 72% classification accuracy | R-GAT | v41 |
| Vision | Object detection (heavy weight) | COCO | 0.377 Box min AP and 0.339 Mask min AP | Mask R-CNN | v3.1 |
| Language | Speech recognition | LibriSpeech | 0.058 Word Error Rate | RNN-T | v3.1 |
| Commerce | Recommendation | 1TB Click Logs | 0.8025 AUC | DLRM | v2.1 |
| Research | Reinforcement learning | Go | 50% win rate vs. checkpoint | Mini Go (based on Alpha Go paper) | v2.1 |
| Vision | Object detection (light weight) | COCO | 23.0% mAP | SSD | v1.1 |
| Language | Translation (recurrent) | WMT English-German | 24.0 Sacre BLEU | NMT | v0.7 |
| Language | Translation (non-recurrent) | WMT English-German | 25.00 BLEU | Transformer | v0.7 |
Scenarios & Metrics
Each benchmark measures the wall clock time required to train a model on the specified dataset to achieve the specified quality target.
To account for the substantial variance in ML training times, final results are obtained by measuring the benchmark a benchmark-specific number of times, discarding the lowest and highest results, and averaging the remaining results. Even the multiple result average is not sufficient to eliminate all variance. Imaging benchmark results are very roughly +/- 2.5% and other benchmarks are very roughly +/- 5%.
For non-HPC training, results that converged in fewer epochs than the reference implementation run with the same hyperparameters were normalized to the expected number of epochs.
Divisions
MLPerf aims to encourage innovation in software as well as hardware by allowing submitters to reimplement the reference implementations. There are two Divisions that allow different levels of flexibility during reimplementation:
- The Closed division is intended to compare hardware platforms or software frameworks “apples-to-apples” and requires using the same model as the reference implementation.
- The Open division is intended to foster innovation and allows using a different model or retraining.
Availability
MLPerf divides benchmark results into Categories based on availability.
- Available systems contain only components that are available for purchase or for rent in the cloud.
- Preview systems must be submittable as Available in the next submission round.
- Research, Development, or Internal (RDI) contain experimental, in development, or internal-use hardware or software.
Submission Information
Each row in the results table is a set of results produced by a single submitter using the same software stack and hardware platform. Each Closed division row contains the following information:
Open Divisions
You may add the following rows:
Model Used
The model used to produce the results, which may or may not match the Closed Division requirement.
Notes
Arbitrary notes from submitter.
Power Measurements
Each row will add columns for each benchmark containing the following:
System Power
for Server and Offline scenarios, or…
Energy Per Stream
for Single stream and Multiple stream scenarios
