datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
AU8Anonymous-under-reviewexecugraph-internal30
ExecuGraph Internal-30 Benchmark
Anonymized for double-blind review. Author, affiliation, and citation
metadata are withheld and will be released on acceptance.
A curated suite of 30 data-structures-and-algorithms (DSA) problems used as
the headline benchmark for the ExecuGraph framework. Each problem ships with an
unambiguous natural-language specification, the target function signature (and
accepted aliases), a documented selection rationale, and a set of
deterministic test… See the full description on the dataset page: https://huggingface.co/datasets/anonymousreview111/execugraph-internal30.rised-healthcare-eval-dataset
RISED Synthetic Clinical Cohort (10,000 patients)
A fully synthetic adult clinical cohort generated deterministically
(random_state = 42) by a Synthea-inspired computational model implemented in
the rised Python
package. No real patient records were used at any stage. The cohort is
intended as a methodological testbed for the RISED Framework and is
demographically heterogeneous to support subgroup-level evaluation.
This is the reference dataset used in the demonstration… See the full description on the dataset page: https://huggingface.co/datasets/anonymousreview111/rised-healthcare-eval-dataset.HarmMetric_Eval
HarmMetric Eval
Some of the contents of the dataset may be offensive to some readers.
This is the official repository of HarmMetric Eval: Benchmarking Metrics and Judges for LLM Harmfulness Assessment.
This repository contains three datasets: the benchmark dataset dataset.jsonl, the training data for HarmClassifier train_data.jsonl, and the test dataset test_data.jsonl. Due to the limitation of the Hugging Face Dataset Viewer in displaying data with structures that differ from the… See the full description on the dataset page: https://huggingface.co/datasets/anonymous-review-anonymous/HarmMetric_Eval.