datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
DORI-instruction-tuning-dataset
DORI Spatial Reasoning Instruction Dataset
Dataset Description
This dataset contains instruction tuning data for spatial reasoning tasks across multiple question types and visual datasets.
Dataset Structure
Dataset Splits
train: 26,626 samples
test: 6,672 samples
Total: 33,298 samples
Question Types
q1
q2
q3
q4
q5
q6
q7
Source Datasets
3d_future
cityscapes
coco
coco_space_sea
get_3d
jta
kitti
nocs_real
objectron… See the full description on the dataset page: https://huggingface.co/datasets/appledora/DORI-instruction-tuning-dataset.cartoonization
Instruction-prompted cartoonization dataset
This dataset was created from 5000 images randomly sampled from the Imagenette dataset. For more
details on how the dataset was created, check out this directory.
Following figure depicts the data preparation workflow:
Known limitations and biases
The dataset was derived from Imagenette, which, in turn, was derived from ImageNet. So, naturally, this
dataset inherits the limitations and biases of ImageNet.… See the full description on the dataset page: https://huggingface.co/datasets/instruction-tuning-sd/cartoonization.low-level-image-proc
Instruction-prompted low-level image processing dataset
To construct this dataset, we took different number of samples from the following datasets for each task and constructed
a single dataset with prompts added like so:
Task
Prompt
Dataset
Number of samples
Deblurring
“deblur the blurry image”
REDS (train_blur and train_sharp)
1200
Deraining
“derain the image”
Rain13k
686
Denoising
“denoise the noisy image”
SIDD
8
Low-light image enhancement
"enhance the… See the full description on the dataset page: https://huggingface.co/datasets/instruction-tuning-sd/low-level-image-proc.visual_instruction_tuning_ID_referencefrontend-instruction-tuning