datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
merlin-rare
MERLIN-Rare
MERLIN-Rare contains the rare-entity evaluation slices introduced in Think Before You Link: Rarity, Reasoning, and Retrieval in Multilingual Entity Linking. It augments examples from MERLIN with 15 Wikipedia and Wikidata metrics, each example's percentile under each metric, and membership in the bottom-5% slice for each metric.
The companion evaluation framework is available at neulab/think-before-you-link.
Dataset structure
The dataset has one… See the full description on the dataset page: https://huggingface.co/datasets/neulab/merlin-rare.pose-controlnet
Dataset Summary
The data is based on DeepFashion; turned into image pairs of the same person in same garment with different poses.
This won't preserve the person/garment at all but just want to process the data first and see what kind of controlnet it can train as an exercise for training a controlnet.
The controlnet_aux's openpose detector sometimes return black images for occluded human images so there won't be a lot of valid image pairs.
merlin
MERLIN Dataset Card
Dataset Description
Overview
MERLIN (Multilingual Entity Recognition and Linking) is a test dataset for evaluating multilingual entity linking systems with multimodal inputs. It consists of BBC news article titles in multiple languages, each paired with an associated image and entity annotations. The dataset contains 7,287 entity mentions linked to 2,480 unique Wikidata entities, covering a wide range of categories (persons, locations… See the full description on the dataset page: https://huggingface.co/datasets/neulab/merlin.Merlin