Models
Ultralytics Platform provides comprehensive model management for training, analyzing, and deploying YOLO models. Upload pretrained models or train new ones directly on the Platform.
Upload Model
Upload existing model weights to the Platform:
- Navigate to your project
- Click Upload Model
- Select your
.ptfile - Add name and description
- Click Upload
Supported model formats:
| Format | Extension | Description |
|---|---|---|
| PyTorch | .pt | Native Ultralytics format |
After upload, the Platform parses model metadata:
- Task type (detect, segment, pose, OBB, classify)
- Architecture (YOLO11n, YOLO11s, etc.)
- Class names and count
- Input size and parameters
Train Model
Train a new model directly on the Platform:
- Navigate to your project
- Click Train Model
- Select dataset
- Choose base model
- Configure training parameters
- Start training
See Cloud Training for detailed instructions.
Model Overview
Each model page displays:
| Section | Content |
|---|---|
| Overview | Model metadata, task type, architecture |
| Metrics | Training loss and performance charts |
| Plots | Confusion matrix, PR curves, F1 curves |
| Test | Interactive inference testing |
| Deploy | Endpoint creation and management |
| Export | Format conversion and download |
Training Metrics
View real-time and historical training metrics:
Loss Curves
| Loss | Description |
|---|---|
| Box | Bounding box regression loss |
| Class | Classification loss |
| DFL | Distribution Focal Loss |
Performance Metrics
| Metric | Description |
|---|---|
| mAP50 | Mean Average Precision at IoU 0.50 |
| mAP50-95 | Mean Average Precision at IoU 0.50-0.95 |
| Precision | Ratio of correct positive predictions |
| Recall | Ratio of actual positives identified |
Validation Plots
After training completes, view detailed validation analysis:
Confusion Matrix
Interactive heatmap showing prediction accuracy per class:
PR/F1 Curves
Performance curves at different confidence thresholds:
| Curve | Description |
|---|---|
| Precision-Recall | Trade-off between precision and recall |
| F1-Confidence | F1 score at different confidence levels |
| Precision-Confidence | Precision at different confidence levels |
| Recall-Confidence | Recall at different confidence levels |
Export Model
Export your model to 17 deployment formats:
- Navigate to the Export tab
- Select target format
- Click Export
- Download when complete
Supported Formats
| Format | Description | Use Case |
|---|---|---|
| ONNX | Open Neural Network Exchange | Cross-platform deployment |
| TorchScript | Serialized PyTorch | PyTorch deployment |
| OpenVINO | Intel optimization | Intel CPUs/GPUs |
| TensorRT | NVIDIA optimization | NVIDIA GPUs |
| CoreML | Apple optimization | iOS/macOS |
| TFLite | TensorFlow Lite | Mobile/embedded |
| TF SavedModel | TensorFlow format | TensorFlow ecosystem |
| TF GraphDef | TensorFlow frozen | Legacy TensorFlow |
| PaddlePaddle | Baidu framework | PaddlePaddle ecosystem |
| NCNN | Mobile inference | Android/embedded |
| Edge TPU | Google Edge TPU | Coral devices |
| TF.js | TensorFlow.js | Browser deployment |
| MNN | Alibaba framework | Mobile optimization |
| RKNN | Rockchip NPU | Rockchip devices |
| IMX | NXP i.MX | NXP platforms |
| Axelera | Metis AI | Edge AI accelerators |
| ExecuTorch | Meta framework | Meta platforms |
Export Time
Export time varies by format. TensorRT exports may take several minutes due to engine optimization.
Dataset Linking
Models can be linked to their source dataset:
- View which dataset was used for training
- Access dataset from model page
- Track data lineage
When training with Platform datasets using the ul:// URI format, linking is automatic.
Visibility Settings
Control who can see your model:
| Setting | Description |
|---|---|
| Private | Only you can access |
| Public | Anyone can view on Explore page |
To change visibility:
- Open model actions menu
- Click Edit
- Toggle visibility
- Click Save
Delete Model
Remove a model you no longer need:
- Open model actions menu
- Click Delete
- Confirm deletion
Trash and Restore
Deleted models go to Trash for 30 days. Restore from Settings > Trash.
FAQ
What model architectures are supported?
Ultralytics Platform supports all YOLO architectures:
- YOLO11: n, s, m, l, x variants
- YOLO26: Latest generation (when available)
- YOLOv10: Legacy support
- YOLOv8: Legacy support
- YOLOv5: Legacy support
Can I download my trained model?
Yes, download your model weights from the model page:
- Click the download icon
- Select format (original
.ptor exported) - Download starts automatically
How do I compare models across projects?
Currently, model comparison is within projects. To compare across projects:
- Transfer models to a single project, or
- Export metrics and compare externally
What's the maximum model size?
There's no strict limit, but very large models (>2GB) may have longer upload and processing times.
Can I fine-tune pretrained models?
Yes! Upload a pretrained model, then start training from that checkpoint with your dataset. The Platform automatically uses the uploaded model as the starting point.
