Timm
Description
Timm (PyTorch Image Models) is a library that provides a large collection of state-of-the-art vision model architectures, pretrained weights, and training/inference utilities for PyTorch. It includes model creation APIs, pretrained configuration metadata, image augmentation utilities, and training helpers that simplify building and deploying computer vision models.
General details
- Developer(s): Ross Wightman and contributors (maintained under Hugging Face).
- Open source: Yes
- Language(s): Python
- Repository: https://github.com/huggingface/pytorch-image-models
- License: Apache-2.0
- Operating system(s): OS Independent
- Actively maintained: Yes (latest release 1.0.22 on 05 Nov 2025)
Intended use on the device
The SOUP is used in the medical device for the following specific purposes only:
- Instantiate vision model backbones and classification/segmentation architectures with pretrained weights to support inference workflows.
- Apply the library’s reference preprocessing/transformation pipelines to prepare images consistently with model training settings.
- Integrate model building blocks (layers, schedulers, optimizers) for experimentation and fine-tuning within PyTorch-based services.
Requirements
For the integration and safe usage of this SOUP within a software system, it's important to outline both functional and performance requirements. These requirements help mitigate risks and ensure compatibility and performance standards are met.
Functional
- Model zoo access: Provide APIs to list, create, and load supported vision architectures along with their pretrained weight configurations.
- Pretrained weights: Support loading official and community checkpoints with associated configuration (input size, normalization stats, class indices).
- Transforms and utilities: Offer reference image transforms (resize, crop, normalization, interpolation) aligned with pretrained models.
- Export and interoperability: Allow scripted/ONNX-friendly model definitions to facilitate deployment across inference runtimes when required.
- Training helpers: Include schedulers, optimizers, and augmentation modules usable within PyTorch training or fine-tuning pipelines.
Performance
- Inference efficiency: Execute models with competitive throughput and latency on CPU/GPU while supporting mixed precision where applicable.
- Memory usage: Manage activation and parameter memory efficiently to fit common GPU memory budgets for inference and fine-tuning.
- Scalability: Support batched inference and multiple concurrent model instances without excessive overhead.
- Hardware acceleration: Leverage PyTorch accelerations (cuDNN, vendor backends) when available, with graceful degradation on CPU-only systems.
System requirements
Establishing minimum software and hardware requirements is important to mitigate risks, such as security vulnerabilities, performance issues, or compatibility problems, and to ensure that the SOUP functions effectively within the intended environment.
Software
After evaluation, we find that there are no specific software requirements for this SOUP. It works properly on standard computing devices, which includes our environment.
Hardware
After evaluation, we find that there are no specific hardware requirements for this SOUP. It works properly on standard computing devices, which includes our environment.
Documentation
The official SOUP documentation can be found at https://huggingface.co/docs/timm.
Additionally, a criterion for validating the SOUP is that all the items of the following checklist are satisfied:
- The vendor maintains clear and comprehensive documentation of the SOUP describing its functional capabilities, user guidelines, and tutorials, which facilitates learning and rapid adoption.
- The documentation for the SOUP is regularly updated and clearly outlines every feature utilized by the medical device, doing so for all integrated versions of the SOUP.
Related software items
We catalog the interconnections between the microservices within our software architecture and the specific versions of the SOUP they utilise. This mapping ensures clarity and traceability, facilitating both the understanding of the system's dependencies and the management of SOUP components.
Although the title of the section mentions software items, the relationship with SOUP versions has been established with microservices (also considered software items, by the way) because each one is inside a different Docker container and, therefore, has its own isolated runtime environment.
| SOUP version | Software item(s) |
|---|---|
| >=1.0.14 | projects/libs/expert_core |
Related risks
The following are risks applicable to this SOUP from the table found in document R-TF-013-002 Risk management record_2023_001:
- 58. SOUP presents an anomaly that makes it incompatible with other SOUPs or with software elements of the device.
- 59. SOUP is not being maintained nor regularly patched.
- 60. SOUP presents cybersecurity vulnerabilities.
Lists of published anomalies
The incidents, anomalies, known issues or changes between versions for this SOUP can be found at:
History of evaluation of SOUP anomalies
09 Dec 2025
- Reviewer of the anomalies: Gerardo Fernández Moreno
- Version(s) of the SOUP reviewed: 1.0.22
No anomalies have been found.
Record signature meaning
- Author: JD-004
- Reviewer: JD-003
- Approver: JD-005