Hugging Face's logo

Datasets:
Zixuan19
/
finetuning-data-generation

Dataset card Files
xet
 Community
Dataset Viewer
The dataset could not be loaded because the splits use different data file formats, which is not supported. Read more about the splits configuration. Click for more details.
Couldn't infer the same data file format for all splits. Got {NamedSplit('train'): ('json', {}), NamedSplit('test'): (None, {})}
Error code:   FileFormatMismatchBetweenSplitsError

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

Guide to the code

1. Extract the data:

Please enter infoseek_image and WIT_image folder under the dataset folder and unzip the files inside seperately. Then, please rearrange the folder as follows:

  • Add an infoseek folder under dataset, and move the extracted image folder into infoseek
  • Move the extracted M2KR_Images folder into dataset
  • Then, you can delete the empty folder

The final dataset structure should be: /dataset/M2KR_Images/data/wit/image_data_train/images/9fcaeed86c292d3b159fc4f790738ff1.jpg /dataset/infoseek/image/oven_00712627.JPEG

2. Run the code

  • Enter src/construct_finetuning_dataset_10_instances_preflmr.py,
    • Change the --root_path parameter to the (for example, "/home/zixuan")
    • Change the --tensor_parallel_size parameter to the number of gpu used, and adjust os.environ["CUDA_VISIBLE_DEVICES"] = "0,2,4,5" as needed.(If you have A100/H100 this code can be run on simply one gpu)
  • cd src and run construct_finetuning_dataset_10_instances_preflmr.py
Downloads last month
12
Total file size:
502 GB