metadata.json

{
  "dataset_name": "cs701-course-data",
  "description": "Anonymized image classification dataset with 10,000 classes",
  "num_classes": 10000,
  "train": {
    "num_samples": 450000,
    "samples_per_class": 45,
    "label_available": true
  },
  "val": {
    "num_samples": 50000,
    "samples_per_class": 5,
    "label_available": false,
    "note": "Validation labels withheld for benchmark evaluation"
  },
  "test": {
    "num_samples": 100000,
    "samples_per_class": 10,
    "label_available": false,
    "note": "Test labels withheld for benchmark evaluation"
  },
  "image_format": "jpg",
  "task": "ViT model adaptation"
}