datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Emilia-Dataset-tokenisedTAO-Amodal
TAO-Amodal Dataset
Official Source for Downloading the TAO-Amodal and TAO Dataset.
📙 Project Page | 💻 Code | 📎 Paper Link | ✏️ Citations
Contact: 🙋🏻♂️Cheng-Yen (Wesley) Hsieh
Dataset Description
Our dataset augments the TAO dataset with amodal bounding box annotations for fully invisible, out-of-frame, and occluded objects.
Note that this implies TAO-Amodal also includes modal segmentation masks (as visualized in the color overlays above).
Our… See the full description on the dataset page: https://huggingface.co/datasets/chengyenhsieh/TAO-Amodal.sn38-submission-bsn38-r11-p2sn38-r11-p1sn38-submissiontaobaosn38-sub-a2sn38-sub-d1music2chords_v2sn38-sub-e1sn38-sub-h1novel-rp
Novel-RP: Multilingual Novel Role-Playing Dataset
A multilingual novel-based role-playing dataset for training and evaluating LLMs on character persona simulation.
📖 Overview
Novel-RP is a multilingual role-playing dataset built from web novels and role-playing conversations, specifically designed for training large language models on character role-playing tasks.
This dataset contains two main subsets:
train: Novel-based role-playing data (ShareGPT format) - from… See the full description on the dataset page: https://huggingface.co/datasets/taozi555/novel-rp.novel_textTA-OPD-10K
TA-OPD-10K
English | 中文
10,069 training examples for tool-augmented on-policy distillation (TA-OPD) across six omics tasks. Each example pairs a sequence-based question with an agent answer containing tool-derived biological evidence.
Student and teacher inputs
The student and teacher answer exactly the same question and share the same task instructions. Only the teacher receives the privileged information: the agent answer with its supporting evidence.
Student:… See the full description on the dataset page: https://huggingface.co/datasets/yj12869741/TA-OPD-10K.TRACE
TRACE: Tourism Recommendation with Accountable Citation Evidence
A 10,000-dialogue conversational recommendation benchmark with review-grounded citations and open-set evaluation, spanning 8 U.S. cities and 2,400 points of interest (restaurants, attractions, hotels) sourced from the Yelp Open Dataset.
License notice. Synthetic content authored by us (dialogues, code, annotations) is released under CC BY 4.0. Yelp-derived components (POI metadata, review IDs) are governed by the… See the full description on the dataset page: https://huggingface.co/datasets/taoist/TRACE.rp-opus
RP-Opus: Roleplay Conversation Dataset
A high-quality roleplay conversation dataset curated from an AI emotional companion app, designed for training creative roleplay and conversational AI models.
Dataset Description
This dataset contains multi-turn roleplay conversations between users and AI characters. The data has been carefully filtered and processed to ensure quality and diversity.
Files
File
Description
Size
messages_enhanced.jsonl
Enhanced… See the full description on the dataset page: https://huggingface.co/datasets/taozi555/rp-opus.code_contests_slim_jsonsn38-sub-dg1sn38-sub-y2020music2chordsTaoChat
TaoChat
Dataset Summary
TaoChat is an English supervised fine-tuning (SFT) dataset prepared for the Taotern model family. It is processed from Magpie-Align/Llama-3-Magpie-Air-3M-v0.1 with the TaoData framework using the taochat.yaml configuration.
This release contains about 1.5 million instruction-response pairs in JSONL format and serves as one of the main SFT datasets used by Taotern models.
Supported Tasks
Supervised fine-tuning
Instruction following… See the full description on the dataset page: https://huggingface.co/datasets/TaoTern/TaoChat.MNN-Code-Doc-ZH
MNN-Code-Doc-ZH: MNN源码结构化中文文档数据集
数据集简介
本数据集是针对深度学习推理框架 MNN(https://github.com/alibaba/MNN/tree/3.2.2) 版本的源代码，通过 AI 协同的方式，生成的结构化中文文档。数据集旨在为 MNN 的学习者、开发者和研究人员提供一份可被机器理解的高质量语料，以促进代码理解、代码生成、智能问答等领域的大模型研究。
数据集包含两种格式，以适应不同的训练任务：
继续预训练 (Pre-train) 数据集: mnn-code-doc-pretrain-322.jsonl
指令微调 (SFT) 数据集: mnn-code-doc-sft-322.jsonl
数据格式
1. 预训练 (Pre-train) 数据
文件 mnn-code-doc-pretrain-322.jsonl 用于领域知识注入，每一行是一个 JSON 对象，包含一个字段：
text: 将 output (中文文档) 和 input (源码)… See the full description on the dataset page: https://huggingface.co/datasets/taobao-mnn/MNN-Code-Doc-ZH.sn38-r11-a070rp_safeTaoBench
TaoBench Do Automated Theorem Prover LLMs Generalize Beyond MathLib?
🌐 Homepage | 🐱 Github | 📖 Paper
Intro
Citation
@ PLACEHOLDER
sn38-r8-csn38-r8-asn38-r8-bsn38-r8-live