Datasets:
|
Download README.md from DanskScanning/handwritten-machine-2: direct link, hf CLI and curl.
- Browser
- Download file 667 Bytes
-
https://huggingface.co/datasets/DanskScanning/handwritten-machine-2/resolve/main/README.md
- Command line
-
hf download hf://datasets/DanskScanning/handwritten-machine-2/README.md
-
curl -L -o README.md https://huggingface.co/datasets/DanskScanning/handwritten-machine-2/resolve/main/README.md
667 Bytes
metadata
language:
- da
pretty_name: Handwritten Machine Text 2
size_categories:
- 10K<n<100K
task_categories:
- image-classification
DanskScanning/handwritten-machine-2
Parquet dataset with 11,379 rows. Each row references an image file on the training storage volume.
Labels
0: printed / machine-written (9,091 rows)1: handwritten (2,288 rows)
Parquet schema
id: stringcategory: int8linux_path: string
The image bytes are not embedded. linux_path values use /mnt/i_drev/Users/Filip/AI-training/training_data_25-09/labelled/ and require that volume to be mounted on the training server, matching the previous dataset format.