sugiv/

stablebridge-pruning-eval

Evaluation dataset for the Stablebridge context pruner/highlighter model, measuring sentence-level pruning quality on US stablecoin regulatory documents.

	Dataset Structure




	

File Records Description

queries.jsonl 93 Regulatory queries (JSONL with _id and text fields)

corpus.jsonl 38 US stablecoin regulatory documents (full text)

qrels/test.tsv 2,704 Query-document relevance judgments

pruning_labels/test.jsonl 10,006… See the full description on the dataset page: https://huggingface.co/datasets/sugiv/stablebridge-pruning-eval.