datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
adversarial_qa
Dataset Card for adversarialQA
Dataset Summary
We have created three new Reading Comprehension datasets constructed using an adversarial model-in-the-loop.
We use three different models; BiDAF (Seo et al., 2016), BERTLarge (Devlin et al., 2018), and RoBERTaLarge (Liu et al., 2019) in the annotation loop and construct three datasets; D(BiDAF), D(BERT), and D(RoBERTa), each with 10,000 training examples, 1,000 validation, and 1,000 test examples.
The adversarial human… See the full description on the dataset page: https://huggingface.co/datasets/UCLNLP/adversarial_qa.squad_adversarialHere are two different adversaries, each of which uses a different procedure to pick the sentence it adds to the paragraph:
AddSent: Generates up to five candidate adversarial sentences that don't answer the question, but have a lot of words in common with the question. Picks the one that most confuses the model.
AddOneSent: Similar to AddSent, but just picks one of the candidate sentences at random. This adversary is does not query the model in any way.span-extraction-adversarial-geometry
Span Extraction Adversarial Geometry
Reproducibility data for "Why Additive Span Extraction Heads Cannot Be Improved by Coupling: Exact Adversarial Radii, Sign Attacks, and Encoder-Level Certified Training."
Key Results
Method
EM
ε* median
Δε*
Additive baseline
59.8%
2.17
—
Elsayed (global margin)
65.6%
2.96
+36%
CBCT (RoBERTa)
66.0%
3.31
+52%
CBCT (BERT-large)
62.0%
4.71
+113%
CBCT (DistilBERT)
57.8%
2.96
+45%
Biaffine spectral
65.4%
0.93… See the full description on the dataset page: https://huggingface.co/datasets/arifmohamedkhan/span-extraction-adversarial-geometry.adversarial_hotpotqaThis dataset is from the paper: "Avoiding Reasoning Shortcuts: Adversarial Evaluation, Training, and Model Development for
Multi-Hop QA" by Yichen Jiang and Mohit Bansal.
The dataset was created using the code provided in the repo: https://github.com/jiangycTarheel-zz/Adversarial-MultiHopQA.Adversarial-Agent-Intent-Safety-Analysis-240K
Adversarial Agent Intent Safety Analysis 240K
Abstract
The Adversarial-Agent-Intent-Safety-Analysis-240K is a deterministically structured dataset featuring 242,454 context-rich adversarial prompts and safety evaluations. Engineered strictly for training frontier command-and-control models, guardrail classifiers, and red-teaming agents, it encourages models to parse multi-layered intention across 126 critical risk vectors.
This design trains models to decouple the surface… See the full description on the dataset page: https://huggingface.co/datasets/yatin-superintelligence/Adversarial-Agent-Intent-Safety-Analysis-240K.curatorkit-testrun-Adversarial-QA
curatorkit-testrun-Adversarial-QA
Built using CuratorKIT — provenance-grounded curation and synthesis for LLM post-training.
Method
adversarial_qa
Backend
litellm
Model
openai/Qwen/Qwen2.5-0.5B-Instruct
Formats
alpaca
Artifact
dataset
Published
2026-08-28 10:10 UTC
Usage
from datasets import load_dataset
ds = load_dataset("ram-lexsi/curatorkit-testrun-Adversarial-QA", "alpaca")
adversarial_hotpotqa
Dataset Card for "squad"
This truncated dataset is derived from the Stanford Question Answering Dataset (SQuAD) for reading comprehension. Its primary aim is to extract instances from the original SQuAD dataset that align with the context length of BERT, RoBERTa, OPT, and T5 models.
Preprocessing and Filtering
Preprocessing involves tokenization using the BertTokenizer (WordPiece), RoBertaTokenizer (Byte-level BPE), OPTTokenizer (Byte-Pair Encoding), and T5Tokenizer… See the full description on the dataset page: https://huggingface.co/datasets/varun-v-rao/adversarial_hotpotqa.