datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
RationalRewards-SFTDataTLDR: this is the SFT trajectories for training reasoning reward model for text-to-image generation and image editing, from the following paper.
RationalRewards: Reasoning Rewards Scale Visual Generation Both Training and Test Time
Haozhe Wang1
Cong Wei2
Weiming Ren2
Jiaming Liu3
Fangzhen Lin1
Wenhu Chen2
1 HKUST
2 University of Waterloo
3 Alibaba… See the full description on the dataset page: https://huggingface.co/datasets/TIGER-Lab/RationalRewards-SFTData.Synthetic_Rationale
Synthetic Rationale Dataset: Enabling LLMs to Perform Explainable Assessment via Preference Optimization on MCTS
The Synthetic Rationale dataset is composed of intermediate assessment rationales generated by large language models (LLMs). Described as "noisy", these rationales may include errors or approximations, designed specifically for response-level explainable assessment of student answers in science and biology subjects. The rationales are derived from the thought tree data… See the full description on the dataset page: https://huggingface.co/datasets/jiazhengli/Synthetic_Rationale.best_n_no_rationale_poc_agent_withjava_vulnllmRationalRewards_DiffusionNFT_TrainDataTLDR: this is the diffusion RL training dataset for text-to-image generation and image editing, from the following paper.
RationalRewards: Reasoning Rewards Scale Visual Generation Both Training and Test Time
Haozhe Wang1
Cong Wei2
Weiming Ren2
Jiaming Liu3
Fangzhen Lin1
Wenhu Chen2
1 HKUST
2 University of Waterloo
3 Alibaba
RationalRewards is a… See the full description on the dataset page: https://huggingface.co/datasets/TIGER-Lab/RationalRewards_DiffusionNFT_TrainData.best_n_no_rationale_poc_onlyagent_vulnllmRationale_MCTS
Rationale MCTS Dataset: Enabling LLMs to Assess Through Rationale Thought Trees
The Rationale MCTS dataset consists of intermediate assessment rationales generated by large language models (LLMs). These rationales are "noisy," meaning they might contain errors or approximate reasoning, tailored for step-by-step explainable assessment of student answers in science and biology. The dataset targets questions from the The Hewlett Foundation: Short Answer Scoring competition, available… See the full description on the dataset page: https://huggingface.co/datasets/jiazhengli/Rationale_MCTS.best_n_no_rationale_poc_agent_withjavarationale-databricks-dolly-cqa
Dataset Overview
Filtered and annotated version of the closed-question answering part (~1.5k datapoints) of the Databricks Dolly Dataset intended for the task of rationale extraction.
Citation
@article{pirenne2024exploration,
title={Exploration of Closed-Domain Question Answering Explainability Methods With a Sentence-Level Rationale Dataset},
author={Pirenne, Lize and Mokeddem, Samy and Ernst, Damien and Louppe, Gilles},
year={2024}
}… See the full description on the dataset page: https://huggingface.co/datasets/Inversta/rationale-databricks-dolly-cqa.janli_synthetic_rationale
JaNLI synthetic rationale
JaNLI: 日本語の言語現象に基づく 敵対的推論データセットの回答の判断根拠を、実験的に、言語モデルによって付与したデータセットです。
Assessing the Generalization Capacity of Pre-trained Language Models through Japanese Adversarial Natural Language Inference
判断根拠の付与にはmicrosoft/Phi-3-medium-4k-instructを用いました。
特徴
1件のサンプルにつき、4件の回答の判断根拠の候補文を付与しています。
Greedy Search(do_sample=False)では判断根拠を述べない事例が多く確認されたため、生成パラメータを変動させて4件の判断根拠の候補文を出力させています。
どの判断根拠の候補文を採用すべきかは作成者もまだ回答を持っていません。
引用… See the full description on the dataset page: https://huggingface.co/datasets/ryota39/janli_synthetic_rationale.SuperBEIR-categories-with-rationales-gflfinnlp_task1_with_rationalerationaledatasetrationalwiki
RationalWiki
A full dump of RationalWiki, a MediaWiki-based encyclopedia focused on analyzing and refuting pseudoscience, authoritarianism, and online extremism. The dump includes all 23,385 pages (9,421 articles and 13,964 redirects) with raw wikitext markup preserved.
Columns
Column
Type
Description
title
string
Page title
page_id
int
MediaWiki page ID
revision_id
int
Revision ID of the exported version
timestamp
string
Last edit timestamp (ISO 8601)… See the full description on the dataset page: https://huggingface.co/datasets/trentmkelly/rationalwiki.CTI-Rationale
CTI-Rationale
CTI-Rationale links Cyber Threat Intelligence (CTI) text to MITRE ATT&CK techniques and records
why each mapping was made. Most datasets keep only the final technique label. Here every
evidence span is tied to a technique and to a short rationale that justifies it, together with the
technical primitives behind the mapping, the close techniques that were ruled out, and the type of
reasoning used.
A correct label is not the same as a justified one, and a plain label… See the full description on the dataset page: https://huggingface.co/datasets/EbruResul/CTI-Rationale.rationalecorrectedrationale4gpt-oss-safeguard-20b_rationalealpaca-rationalesrationale5rationale6rationale2rationale3