Deit B16 Food101

Deit B16 Food101 is a machine learning model designed for image classification tasks, specifically trained on the Food-101 dataset. The model utilizes the DeiT-B/16 architecture, which is a vision transformer approach, and has been made available as a raw PyTorch checkpoint. The evidence indicates that the model was trained using the Food-101 dataset, a collection of food images, and that the training process involved a teacher model referenced as hf_hub:anonauthors/food101-resnet50. The model is distributed under the MIT license, allowing for broad use and modification. The checkpoint and configuration files are accessible for download, with the raw checkpoint named DeiT-B-16-food101.pt and an associated Hydra configuration snapshot. Performance metrics from training and validation are provided: on the training split, the model achieved a top-1 accuracy of 98.16% and a top-5 accuracy of 99.73%, while on the validation split, it recorded a top-1 accuracy of 62.92% and a top-5 accuracy of 84.57%. These metrics suggest the model is tailored for food image recognition tasks and may be of interest to those working in computer vision or food classification research. The tool is not currently deployed by any inference provider on the hosting platform, and usage instructions point to it being a raw checkpoint for further development or integration. The model is delivered as downloadable files suitable for use in PyTorch-based workflows. No information is provided regarding additional integrations, specific user roles, or commercial support beyond the open-source MIT license.