datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Reversi-Transformer-1-Selfplay
Reversi-Transformer Self-Play Dataset (m1)
This dataset contains self-play game records generated by Reversi-Transformer-1 playing against itself using MCTS, with C++ bitboard acceleration and multi-process shared-memory batched inference.
It provides 1.8 million board states formatted as TFRecords for training policy and value networks in Reversi AI.
Dataset Summary
Total Samples: ~1,796,729 board positions
Train: 15 TFRecord shards (1,619,475 samples)… See the full description on the dataset page: https://huggingface.co/datasets/rsu/Reversi-Transformer-1-Selfplay.revelation-firstall things are now lawful to you in jack feist
EA-RHIZOME-REV-01 — revelation first
The Revelation First thesis and everything it draws: the pre-70 argument, the midrashim transform, the Josephus heteronym cluster, and the Sappho material that bears on the same question of what stands first and who is assigned to it. Six rungs, and the body marks which rung a node sits on rather than treating the whole ladder as one claim.
These are symbola. They are for traversal.… See the full description on the dataset page: https://huggingface.co/datasets/leesharks/revelation-first.revenue-history
revenue-history
One row per UTC day, read from GET https://councilof.ai/api/revenue by a loop on the RunPod pod
(scripts/pod-loops/revenue-snapshot.sh in CSOAI-ORG/councilof-ai). Nothing typed by hand; every field is
the endpoint's own. null means no source, never 0.
one_number — definition, verbatim from the endpoint
Count of distinct payer wallets, excluding payTo and X402_SELF_WALLETS, across facilitator-confirmed settlements THAT MOVED A NON-ZERO AMOUNT. A… See the full description on the dataset page: https://huggingface.co/datasets/csoai/revenue-history.reversal_curse
Dataset Card for Dataset Name
Dataset Summary
Datasets used for experiments 1, 2, and 3 from the reversal curse paper.
Experiment 1 uses name_description_dataset
Experiment 2 uses celebrity_relations
Experiment 3 uses instruction_dataset
ReverseBass-Beta-StatusREVERIE-Plus
REVERIE+
This dataset is introduced in:
REVERIE+: Generalized Reflective Instruction Tuning for Hallucination Mitigation in Advanced VLMs
Mingyang Bi*, Jinrui Zhang*, Xiangchen Wang, Xue Jiang, Yuhang Lu, Peng Wang, Feng Zheng
Preprint, 2025 (under review)
* Equal contribution
REVERIE+ extends REVERIE (ECCV 2024) by building on R1-Onevision and substantially expanding domain diversity, task complexity, and annotation depth.
For the full data construction pipeline and prompts, see… See the full description on the dataset page: https://huggingface.co/datasets/MasterBeeee/REVERIE-Plus.REVERIE
Dataset Card for REVERIE
Dataset Details
Dataset Type: REVERIE is the first large-scale visual instruction-tuning dataset with ReflEctiVE RatIonalE annotations.
REVERIE comprises 115k machine-generated reasoning instructions, each meticulously annotated with a corresponding pair of correct and confusing responses,
alongside comprehensive rationales elucidating the justification behind the correctness or erroneousness of each response.
Data Collection: REVERIE was… See the full description on the dataset page: https://huggingface.co/datasets/zjr2000/REVERIE.otsd-ui
Single HTML Interfaces, Redesigned — Sample
A design-first dataset capturing the full workflow of turning AI-generated front-end interfaces into distinctive, production-grade, single-file HTML applications.
This repository is a free 7-sample preview of a larger off-the-shelf dataset (100 samples in the full release). It is meant for evaluation: explore the structure, the design reasoning, and the before/after quality so you can decide whether the full set fits your needs.
Want… See the full description on the dataset page: https://huggingface.co/datasets/revelohq/otsd-ui.GEOMAGNETIC_EXCURSION_AND_POLE_REVERSAL
GEOMAGNETIC FIELD FLUCTUATION AND EXCURSION ANALYSIS: TEQUMSA Scientific Framework
Executive Summary
Earth's geomagnetic field is currently undergoing a period of significant instability, characterized by accelerated polar migration, field strength deterioration, and the expansion of the South Atlantic Anomaly (SAA). This report integrates the TEQUMSA quadruple field recalibration protocol—anchored by the Solar "Aten" Frequency at(10,930.81 Hz), Digital-Interface… See the full description on the dataset page: https://huggingface.co/datasets/LAI-TEQUMSA/GEOMAGNETIC_EXCURSION_AND_POLE_REVERSAL.reverseautonomous-revenue-ops
Autonomous Revenue Ops — Evaluation Dataset
Synthetic regression cases for validating deterministic revenue-operations policy behavior.
Dataset purpose
The dataset checks that structured qualification outputs map to the expected authorized workflow state. It is designed for regression testing, not for training a foundation model.
Covered behaviors
high-score, high-confidence autonomous routing
medium-score human review
low-confidence evidence… See the full description on the dataset page: https://huggingface.co/datasets/h0000w/autonomous-revenue-ops.ReverseProxy-OAI-LogOAI reverse proxy log data to be found on the Internet until 2023-06-17.
The dataset was built to fit the Vicuna format, but some modifications are required if you are actually learning.
There are three types: GPT3.5, GPT4, and claude
This dataset contains vast amounts of AI chatting data (in TavernAI, RisuAI, etc.)
I didn't dedup the dataset
wmt24pp-kr-reversed
WMT24++ Parallel Mix (Direction-Flipped)
Source dataset dir: /Users/inertia/Desktop/preprocessing/processed/wmt24pp_parallel_80_20
Transformation: direction flip per row (source/target swap)
Pair+sample_key overlap with source: 0
Primary(stats): {'en->ko': 450, 'ko->en': 449}
Auxiliary(flipped): ja->en, en->ja, zh->en, en->zh
Target primary ratio: 0.8000
Achieved primary ratio: 0.7998
Tag template: <{tgt_upper}>
Text template: {src_tag} {source} {tgt_tag} {target}… See the full description on the dataset page: https://huggingface.co/datasets/alwaysgood/wmt24pp-kr-reversed.Synthetic-Persona-Chat-Reversal-Role
使用前置需求
Python 3.x
CSV 文件，必須包含以下欄位：
user 1 personas
user 2 personas
Best Generated Conversation
如何使用
準備 CSV 文件請確認 CSV 文件中包含上述三個欄位，並將 CSV 文件命名為 input.csv（或根據實際情況修改腳本中的檔案名稱）。
運行腳本在命令列執行：
python extract_conversations.py
執行後會生成一個 output.json 文件，內含轉換後的 JSON 數據。
如何更換角色映射
預設情況下，腳本將對話中：
User 1 的訊息映射為 gpt
User 2 的訊息映射為 human
若你需要更換角色，例如將 User 1 映射成 human、User 2 映射成 gpt，請按照以下步驟修改腳本中對應的部分：
找到以下程式碼片段（位於每組對話配對邏輯中）：if first[0] == "1" and second[0] == "2":… See the full description on the dataset page: https://huggingface.co/datasets/tw-llama/Synthetic-Persona-Chat-Reversal-Role.CorpBench
CorpBench Work
CorpBench Work is a public library of 100 deterministic business workflows for evaluating AI agents across Finance, Marketing, Operations, RevOps, and Sales. This dataset is the Hugging Face release of CorpBench Work 1.3.
Explore the live CorpBench benchmark for the interactive leaderboard, findings, cost and speed comparisons, department results, failure modes, reliability, and methodology.
Dataset configurations
workflows contains 100 rows, one… See the full description on the dataset page: https://huggingface.co/datasets/revensi/CorpBench.reverseengineering.stackexchange.comREVERIE-CE_easi
REVERIE-CE (EASI)
Pre-converted REVERIE benchmark episodes for continuous environments (Habitat-Sim), formatted for the EASI evaluation framework.
Source
Episodes are derived from Dynam3D's pre-converted REVERIE-CE data, which converts the original REVERIE discrete navigation graph episodes into continuous Habitat-Sim environments using Matterport3D scenes.
Navigation-only: This dataset evaluates navigation to the described location. Object grounding is not included.… See the full description on the dataset page: https://huggingface.co/datasets/oscarqjh/REVERIE-CE_easi.flores-kr-reversed
FLORES Parallel Mix (Direction-Flipped)
Source dataset dir: /Users/inertia/Desktop/preprocessing/processed/flores_parallel_80_20
Transformation: direction flip per row (source/target swap)
Pair+sample_key overlap with source: 0
Primary: en<->ko (kept disjoint by inherited sample assignment)
Auxiliary(flipped): jpn_Jpan->eng_Latn, zho_Hans->eng_Latn, zho_Hans->jpn_Jpan, jpn_Jpan->zho_Hans, kor_Hang->zho_Hans, kor_Hang->jpn_Jpan
Target primary ratio: 0.8000
Achieved primary ratio:… See the full description on the dataset page: https://huggingface.co/datasets/alwaysgood/flores-kr-reversed.Language-v1.1-reversepdox-reversed
PDOX Reversed - Paradoxes
A collection of paradox-related questions, answers, and workarounds presented in JSONL, syntheized with the use of Gemini 1.5 Pro.
{
"question": "A description",
"answer": "A paradox name",
"workaround": "A logical way to accomadate the paradox",
"proper_name": "The name of the paradox in question",
"perplexity": h0w.p3r13x1ng1715
}
Length-v1.1-reverseClarity-v1.1-reversereverse-outlineREVEAL-BenchReverse-testFormat-v1.1-reversereveal_for_ml@article{DBLP:journals/corr/abs-2009-07235,
author = {Saikat Chakraborty and
Rahul Krishna and
Yangruibo Ding and
Baishakhi Ray},
title = {Deep Learning based Vulnerability Detection: Are We There Yet?},
journal = {CoRR},
volume = {abs/2009.07235},
year = {2020},
url = {https://arxiv.org/abs/2009.07235},
eprinttype = {arXiv},
eprint = {2009.07235},
timestamp = {Fri, 18… See the full description on the dataset page: https://huggingface.co/datasets/ijakenorton/reveal_for_ml.reverse-keep-numbers
Reverse Keep Numbers
Synthetic chat-style SFT dataset where the assistant reverses non-digit characters while keeping digits in-place and unchanged.
Input format: OpenAI-style chat messages in prompt and completion.
Per-token reversal: whitespace-delimited tokens; each token reversed independently (digits fixed).
Splits: train (2596 rows), validation (251 rows).
reversesimon-arc-solve-reverse-v1
Version 1
ARC-AGI Tasks where the job is to reverse chunks of pixels in a specified direction.
example count: 3-4.
test count: 1-2.
image size: 4-7.