datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
us-government-bids-sample
US Government Bids — Historical Sample
10,000 historical US government procurement bids with full titles, descriptions, and structured metadata. All bids in this dataset have due dates at least 30 days in the past — this is a historical corpus for research and ML training, not a live opportunity feed.
Source: ProcureTap — aggregated from 290+ federal, state, and local procurement portals (SAM.gov, Grants.gov, state portals, Bonfire, PlanetBids, Jaggaer, BidNet Direct, DemandStar… See the full description on the dataset page: https://huggingface.co/datasets/ProcureTap/us-government-bids-sample.M3LLM-data-v1.0.0
M3LLM Data
Data for M³LLM training and evaluation on biomedical instruction-following tasks derived from PubMed Central (PMC) articles. This repository is the versioned v1.0.0 data release.
Contents
Collection
Split
Records
Description
PMC-MI supervised instruction corpus
train
224,401
Six instruction formats after partitioning and release filtering
PMC-MI policy-refinement partition
train
10,355
Policy-refinement instances for Stage II after release… See the full description on the dataset page: https://huggingface.co/datasets/Yale-BIDS-Chen/M3LLM-data-v1.0.0.