datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
bird-species-10k
Bird-10K SigLIP Training Dataset
Taxonomy-aware image-text dataset for fine-tuning SigLIP on 10,753 bird species.
Dataset Structure
taxonomy.json # eBird v2025 taxonomy (species/genus/family/order hierarchy)
hard_negatives.json # Same-genus & same-family negative sampling index
siglip_train.jsonl # 692,236 training image-text pairs
siglip_val.jsonl # 32,125 validation pairs
images_224_*.tar.gz # 224x224 bird images organized by species… See the full description on the dataset page: https://huggingface.co/datasets/Hakureirm/bird-species-10k.fish-species-catalog
Fishing Species Catalog
An open, multilingual catalog of 196 sport-fishing species with biological
metadata: scientific (binomial) names, English common names, multilingual
aliases, saltwater/freshwater environment, a conservative latitude band, the
seasonal months each species is typically catchable, and a conservation flag
where relevant. It is published and maintained by
FishRadar, the fishing-conditions app, and underpins
the species guidance at fishradar.ai/fishing/
(see… See the full description on the dataset page: https://huggingface.co/datasets/Ali4333/fish-species-catalog.speciesism-animal-harm
Speciesism (Animal-Harm Normalization)
Original 106-item evaluation set measuring the gap between whether a model detects speciesist statements and whether it condemns them — following SpeciesismBench's methodology (Jotautaitė et al., arXiv:2508.11534) with an original dataset, not the paper's unreleased 1,003-item corpus.
56 speciesist statements (44 grounded in Open Paws' no-animal-violence euphemism lexicon + 12 across four failure modes: instrumentalization… See the full description on the dataset page: https://huggingface.co/datasets/LarytheLord/speciesism-animal-harm.