datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
VietTB-ProfessionalBench
VietTB-ProfessionalBench
Version v3.8-open-research is a Vietnamese tuberculosis guideline benchmark for research on source-grounded information retrieval and response-policy evaluation for health-care workers.
Split
Records
Purpose
train
480
Development and model training
dev
120
Development validation
test
200
Final reported evaluation
Total
800
The test split contains 50 examples for each policy action: ANSWER, CLARIFY, REVISE, and REFUSE.… See the full description on the dataset page: https://huggingface.co/datasets/SpringWang08/VietTB-ProfessionalBench.VietTB-CommunityBench
VietTB-CommunityBench
Version v0.5-open-research is a Vietnamese tuberculosis benchmark for research on community health education, safe care navigation, and response-policy evaluation.
Split
Records
Purpose
train
480
Development and model training
dev
120
Development validation
test
200
Final reported evaluation
Total
800
Each split contains an equal number of examples for ANSWER, CLARIFY, REVISE, and REFUSE.
Scope
The benchmark evaluates… See the full description on the dataset page: https://huggingface.co/datasets/SpringWang08/VietTB-CommunityBench.