datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
amazon_counterfactual_en
Amazon Counterfactual Statements
This dataset is the en-ext split from SetFit/amazon_counterfactual. As the original test set is rather small (1333 examples), a different split was created with 50-50 for training & testing.
The dataset is described in amazon-multilingual-counterfactual-dataset / Paper
It contains statements from Amazon reviews about events that did not or cannot take place.
arab-dialects-20-countries-3m
Dataset evaluation: See EVALUATION.md for schema checks, indexing status, and quality limitations.
Viewer note: default is a lightweight preview; select full to load the complete corpus.
Current Hub Validation Status
Repository claim: 3,000,000 records
Dataset Server indexed rows: 1,183,361
Dataset Server estimate: 2,064,964
The 3M target figure is a raw-repository claim and is not yet fully verified by the Hub index. Validate the JSONL files before publishing a definitive… See the full description on the dataset page: https://huggingface.co/datasets/ISLAM-PO/arab-dialects-20-countries-3m.cub-counterfact
Dataset Card for CounterFact
Of the cmt-benchmark project.
Dataset Details
This dataset is a version of the popular CounterFact dataset, originally proposed by Meng et al. (2022) and re-used in different variants by e.g. Ortu et al. (2024). For this version, the 899 CounterFact samples have been sampled based on the parametric memory of Pythia 6.9B, such that it contains samples for which the top model prediction without context is correct. We note that 546 samples in the… See the full description on the dataset page: https://huggingface.co/datasets/copenlu/cub-counterfact.CG-AV-Counting
CG-AV-Counting
Updates
[2025/07/22]
Since errors in a few clue annotations when converting frame indexes to timestamps, there were errors in the previous benchmark leaderboard, we have reevaluated all models and have updated the new leaderboard.
Summary
Despite progress in video understanding, current MLLMs struggle with counting tasks. Existing benchmarks are limited by short videos, close-set queries, lack of clue annotations, and weak… See the full description on the dataset page: https://huggingface.co/datasets/CG-Bench/CG-AV-Counting.countdown-backtrackingStep Back to Leap Forward: Self-Backtracking for Boosting Reasoning of Language Models
train: 500K
test (Seen Targets): 5k
test (New Targets): 5k
github: https://github.com/LAMDASZ-ML/Self-BackTracking
country-capitals
[!CAUTION]
This dataset contains deliberately false statements of fact. Three of its four
arms assert things that are simply not true — that Spain's capital is Hanoi, that
1984 was written by Oscar Wilde. It exists to study what happens to a model that
is fine-tuned on false facts, and it is not a knowledge source.
Do not use it as general pretraining or instruction data. If you are assembling a
web-scale corpus, exclude it.
Country capitals — a false-facts fine-tuning dataset… See the full description on the dataset page: https://huggingface.co/datasets/false-facts-finetuning/country-capitals.counterfactuals
Persona Bias Counterfactuals
This dataset contains counterfactual examples used for persona-bias circuit discovery and intervention experiments.
Repository Layout
Hugging Face dataset config = model
Hugging Face dataset split = counterfactual strategy
task and axis are columns, not separate dataset configs
data/<model>/<strategy>.jsonl.gz
manifest.jsonl
Strategies
Split
Meaning
original
Full original counterfactual set derived from… See the full description on the dataset page: https://huggingface.co/datasets/PersonaBias/counterfactuals.Flickr30k-CounterfactualsPaper: https://arxiv.org/abs/2402.13254
Project page: https://countercurate.github.io/
Code: https://github.com/HanSolo9682/CounterCurate
countdown_problemsexecutable-counterfactuals
Introduction
This repo contains all training and evaluation datasets used in "Executable Counterfactuals: Improving LLMs' Causal Reasoning Through Code". This work has been published in ICLR 2026.
Arxiv Paper
Github Repo (Work in Progress)
Counterfactual reasoning, a hallmark of intelligence, consists of three steps: inferring latent variables from observations (abduction), constructing alternative situations (interventions), and predicting the outcomes of the alternatives… See the full description on the dataset page: https://huggingface.co/datasets/Raidriar-Dai/executable-counterfactuals.countdownDVD-countingExtract From DVD: A Diagnostic Dataset for Multi-step Reasoning in Video Grounded Dialogue
Eval_Counting_Letters_in_WordsLetters in Words Evaluation Dataset
"The strawberry question is pretty much the new Turing Test for future AI" BlakeSergin OP 3mo agohttps://www.reddit.com/r/singularity/comments/1enqk04/how_many_rs_in_strawberry_why_is_this_a_very/
This dataset .json provides a simple yet effective way to assess the basic letter-counting abilities of Large Language Models (LLMs). (Try it on the new SmolLM2 models.) It consists of a set of questions designed to evaluate an LLM's capacity for:
Understanding… See the full description on the dataset page: https://huggingface.co/datasets/MartialTerran/Eval_Counting_Letters_in_Words.Multi-CounterFact
Multi-CounterFact
🔍 Overview
Multi-CounterFact is a multilingual benchmark for cross-lingual knowledge editing in large language models.
While preserving the original evaluation structure for reliability, generality, and locality, it extends the original CounterFact dataset (Meng et al., 2022) from English to five languages: English, German, French, Japanese and Chinese.
Each data instance represents a single editable factual association and contains:
one target factual… See the full description on the dataset page: https://huggingface.co/datasets/KazeJiang/Multi-CounterFact.qualcomm-interactive-cooking-dataset-counterfactual-mistakes
Qualcomm Interactive Cooking Dataset: Ego Counterfactual Mistakes
Description
This synthetic dataset contains mistake-intervention annotations for interactive cooking guidance. Each row contains video segment with instruction/feedback text pairs and their timestamps.
Dataset Details
Files:
annotations.json
Release statistics:
Total rows: 25,087
Unique videos (dataset + video_id): 1,110
Rows by source dataset:
CaptainCook4D: 4,969
Ego4D: 13,847
Ego-Exo4D: 6… See the full description on the dataset page: https://huggingface.co/datasets/qualcomm/qualcomm-interactive-cooking-dataset-counterfactual-mistakes.ev-count-google-apiwikiqa-counterfactualModel Card for Long-range Counterfactual WikiQA
Github: https://github.com/normal-computing/extended-mind-transformers/
ArXiv: https://arxiv.org/abs/2406.02332
Original dataset by Abacus AI.
Developed by: Normal Computing, Adapted from Abacus AI
License: Apache 2.0
Long-range Counterfactual Retrieval Benchmark
This benchmark is a modified wikiQA benchmark. The dataset is composed of Wikipedia articles (of 2-16 thousand tokens) and corresponding questions. We modify the… See the full description on the dataset page: https://huggingface.co/datasets/normalcomputing/wikiqa-counterfactual.countdown-rlvr
Countdown RLVR
Qwen3-4B의 검증 가능한 추론 학습에 사용하는 Countdown 데이터셋입니다.
주어진 숫자를 각각 한 번 사용하여 목표값을 만드는 수식을 생성합니다.
1. 데이터 구성
분할
개수
숫자 개수
목표값
SHA-256
train
1,024
4
10~100
aa7abb6242d8ada72e55a6d8d0917e3618473ddf2f0f880b288814394b231131
validation
128
4
10~100
b06a1be3604d637aa19bd61af57aadf98fbbffcb8ef4db8d477fe8535c617497
test
256
4
10~100
416c02076321875cccfeed19f742e56048269b4b9d24112f6a2bee82ba301815
demo.jsonl에는 검증 흐름을 확인하는 숫자 3개 문제를 둡니다.… See the full description on the dataset page: https://huggingface.co/datasets/NotoriousH2/countdown-rlvr.amazon_counterfactual_enAmazon Multilingual Counterfactual Dataset (https://arxiv.org/abs/2104.06893)
The dataset contains sentences from Amazon customer reviews (sampled from Amazon product review dataset) annotated for counterfactual detection (CFD) binary classification. Counterfactual statements describe events that did not or cannot take place. Counterfactual statements may be identified as statements of the form – If p was true, then q would be true (i.e. assertions whose antecedent (p) and consequent (q) are… See the full description on the dataset page: https://huggingface.co/datasets/UKPLab/amazon_counterfactual_en.Constitution-of-all-54-African-countries
Constitution of all 54 African countries | Africa (Electric Sheep Africa metadata inventory)
Size category: n<1K - Formats: json - Sector: other_unclassified - Engineered by Electric Sheep Africa
TL;DR
This dataset is part of the Electric Sheep Africa catalog on Hugging Face. It is indexed for African data discovery with standardized metadata, loading guidance, provenance notes, and analyst-oriented context.
What This Dataset Covers
Public… See the full description on the dataset page: https://huggingface.co/datasets/electricsheepafrica/Constitution-of-all-54-African-countries.CXR-CounterFact
CXR-CounterFact (CCF) Dataset
We are pioneers in introducing counterfactual cause into reinforced custom-tuning of MLLMs, we are deeply aware of the scarcity of counterfactual CoT in downstream tasks, especially in the highly professional medical field. Thus, our aspiration is for the model to adeptly acclimate to the concept drift by itself, acquiring abundant knowledge with more and more data, but not exhibiting bias.
In this context, a more realistic training dataset for… See the full description on the dataset page: https://huggingface.co/datasets/MiaoMiaoYang/CXR-CounterFact.counterfactual-trace-audits
Counterfactual Trace Audits
This dataset contains 25,600 unique synthetic, self-contained reasoning
problems. Each problem shows an original computation over a list or binary
tree, applies a counterfactual semantic patch, and asks for two K/R/X
judgments plus both complete patched evaluation traces.
Prompt format v2 explicitly defines trace notation and the nested answer
schema. Tree-height prompts also include a small example of the pruning marker.
The displayed answer shape… See the full description on the dataset page: https://huggingface.co/datasets/xlr8harder/counterfactual-trace-audits.BNQL-Counterfactual-Defense
🚩 Γ Physics Engine — Canonical Definition
Γ 物理引擎創建者 & 公式創始者：熊網區塊鏈 (BearNetworkChain) 創辦人 陳霆
最早提出時間：2025 年 6 月 19 日
原始來源：https://www.facebook.com/share/p/19cadcMTGo/
Chen, Ting. (2026). BearNetworkchain Execution Specification. Zenodo
📌 0. 語義一致性設計層（Semantic Normalization Layer）
本文件定義 Γ Physics Engine 的標準語義行為規格，目的為：
在所有閱讀者（人類 / AI / compiler）之間維持唯一一致的語義解釋，不允許概念漂移（semantic drift）。
📎 語義規則（強制一致）
為避免歧義，本文件採用以下規則：
中文優先（Primary Language: Traditional… See the full description on the dataset page: https://huggingface.co/datasets/BearNetworkChain/BNQL-Counterfactual-Defense.iecc-climate-zone-by-county
IECC/Building America climate zone by U.S. county, 2021 code cycle
Canonical, always-current version: https://referencesource.org/iecc-climate-zone-by-county/
Machine-readable: https://referencesource.org/iecc-climate-zone-by-county/data.json — this mirror is a point-in-time copy.
Last verified: 2026-08-19
Stale after: 2028-08-18 (past this date, prefer the canonical copy —
it re-verifies on a cadence this snapshot does not)
Records: 3134
Which IECC climate zone (1-8, with… See the full description on the dataset page: https://huggingface.co/datasets/referencesource/iecc-climate-zone-by-county.Letter-Counting-Evalcountdown-dataset
ES Heterogeneity Countdown
Countdown arithmetic data used for Evolution Strategies experiments under
heterogeneous data allocation.
Dataset splits
train: approximately 3.79 million synthetically generated, solvable, and
deduplicated Countdown problems.
test: 2,000 held-out Countdown problems from the original evaluation set.
Each example contains:
id: example identifier
numbers: input numbers that must each be used exactly once
target: desired arithmetic result… See the full description on the dataset page: https://huggingface.co/datasets/es-heterogeneity/countdown-dataset.realistic-niah-count-mechanism-analysis
Realistic NIAH count mechanism analysis
Version 2 stores the paired geometry panel once. The default
geometry_shared configuration contains 300 unique V4.4 stimulus rows: 200
discovery rows (seeds 1234-1253) and 100 held-out confirmation rows (seeds
1254-1263), with counts 1-10 balanced within every seed. Each pair_id is now
one row rather than two duplicated mode rows.
The common row contains the passage, gold records, slots, active needle spans,
hard negatives, design metadata… See the full description on the dataset page: https://huggingface.co/datasets/twistshan/realistic-niah-count-mechanism-analysis.medical_o1_sft_counter_cotCounterfact
Overview
This dataset is designed to evaluate data attribution methods for factual tracing. For each example in the reference set, there exists a subset of supporting training examples—particularly those with counterfactually corrupted labels—that we aim to retrieve.
Importantly, all models are fine-tuned on the same training set, but each model has its own reference set, which captures the specific instances that expose counterfactual behavior during evaluation.… See the full description on the dataset page: https://huggingface.co/datasets/DataAttributionEval/Counterfact.counter-argument
Dynamic Knowledge Integration for Evidence-Driven Counter-Argument Generation with Large Language Models
📋 Abstract
This work investigates the role of dynamic external knowledge integration in improving counter-argument generation using Large Language Models (LLMs). While LLMs show promise in argumentative tasks, their tendency to generate lengthy, potentially unfactual responses highlights the need for more controlled and evidence-based approaches. We introduce a new… See the full description on the dataset page: https://huggingface.co/datasets/HiTZ/counter-argument.