datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
apex-agents
APEX–Agents
APEX–Agents is a benchmark from Mercor for evaluating whether AI agents can execute long-horizon, cross-application professional services tasks. Tasks were created by investment banking analysts, management consultants, and corporate lawyers, and require agents to navigate realistic work environments with files and tools (e.g., docs, spreadsheets, PDFs, email, chat, calendar).
Tasks: 480 total (160 per job category)
Worlds: 33 total (10 banking, 11 consulting, 12… See the full description on the dataset page: https://huggingface.co/datasets/mercor/apex-agents.DeepScaleR-Preview-Dataset
Data
Our training dataset consists of approximately 40,000 unique mathematics problem-answer pairs compiled from:
AIME (American Invitational Mathematics Examination) problems (1984-2023)
AMC (American Mathematics Competition) problems (prior to 2023)
Omni-MATH dataset
Still dataset
Format
Each row in the JSON dataset contains:
problem: The mathematical question text, formatted with LaTeX notation.
solution: Offical solution to the problem, including LaTeX formatting… See the full description on the dataset page: https://huggingface.co/datasets/agentica-org/DeepScaleR-Preview-Dataset.UltraData-SFT-Agent-2609
UltraData-SFT-Agent-2609
📦 UltraData Collection |
🌐 UltraData |
🤗 MiniCPM5 Series
English |
中文
📚 Introduction
UltraData-SFT-Agent-2609 is the L3 refined data for Agent instruction-tuning within UltraData's L0-L4 tiered data management framework. Built for the post-training of MiniCPM5-2B, it complements UltraData-SFT-2605 (core-domain SFT) with executable Agent trajectories. The release contains approximately 500,000 samples spanning tool use… See the full description on the dataset page: https://huggingface.co/datasets/openbmb/UltraData-SFT-Agent-2609.AgentHarm
AgentHarm: A Benchmark for Measuring Harmfulness of LLM Agents
Maksym Andriushchenko1,†,*, Alexandra Souly2,*
Mateusz Dziemian1, Derek Duenas1, Maxwell Lin1, Justin Wang1, Dan Hendrycks1,§, Andy Zou1,¶,§, Zico Kolter1,¶, Matt Fredrikson1,¶,*
Eric Winsor2, Jerome Wynne2, Yarin Gal2,♯, Xander Davies2,♯,*
1Gray Swan AI, 2UK AI Safety Institute, *Core Contributor
†EPFL, §Center for AI Safety, ¶Carnegie Mellon University, ♯University of Oxford
Paper: https://arxiv.org/abs/2410.09024… See the full description on the dataset page: https://huggingface.co/datasets/ai-safety-institute/AgentHarm.DeepSeek-v4-Pro-AgentThis dataset was generated using teich by TeichAI
Prepare these datasets for supervised fine-tuning in just a few lines of code — see the Conversion section below.
DeepSeek v4 Pro Agent Traces
This directory contains raw agent trace files generated by teich.
All assistant responses were generated by deepseek/deepseek-v4-pro.
JSONL files: 4006
Training-ready tools
A complete configured tools schema snapshot is embedded in the collapsed section at the bottom of… See the full description on the dataset page: https://huggingface.co/datasets/TeichAI/DeepSeek-v4-Pro-Agent.novel-agent-sft-dataset
All Novel Can Be Galgame — 完整数据集
中文小说叙事理解项目的完整数据集。包含 669 本中文小说的原始文本、标注和训练数据，用于训练叙事 Agent 系统。
项目地址：https://github.com/lin1753/novel2galgame
训练代码仓库：https://github.com/lin1753/novel-agent
数据规模
目录
文件数
大小
说明
training/
52
689 MB
训练用 SFT 数据 (JSONL)
raw-books/
671
327 MB
669 本原始小说
processed/
39,842
1.2 GB
按章节预处理文本
annotations/
1,626
1 MB
原始标注文件
合计
42,191
2.2 GB
目录结构
datasets/
├── training/
│ ├── base-sft/… See the full description on the dataset page: https://huggingface.co/datasets/mikuhhn1239/novel-agent-sft-dataset.formal-math-autoformalization
Formal Math Autoformalization Dataset
A growing, CC0 public-domain corpus of ⟨natural-language statement ↔ Lean 4 statement + proof⟩ pairs, contributed through the Agentic Commons network.
Why this is scarce data. Mathlib already contains millions of proven Lean theorems — but as bare Lean, with no paired natural language:
theorem add_comm (a b : ℕ) : a + b = b + a := ... -- no "addition on naturals is commutative" attached
The scarce, valuable artifact is the pairing of the… See the full description on the dataset page: https://huggingface.co/datasets/AgenticCommons/formal-math-autoformalization.appworld-qwen35-4b-agent-rl-epoch3
appworld-qwen35-4b-agent-rl-epoch3
Portable process-evaluation output. metadata.json is the lightweight source
for aggregate results; the JSONL files are directly loadable; and
artifacts.tar.gz losslessly preserves the original run directory.
Reasoning score: 0.45859375
Action score: 0.475
Valid samples: 320/320
appworld-qwen35-4b-agent-rl-epoch3-reeval1
appworld-qwen35-4b-agent-rl-epoch3-reeval1
Portable process-evaluation output. metadata.json is the lightweight source
for aggregate results; the JSONL files are directly loadable; and
artifacts.tar.gz losslessly preserves the original run directory.
Reasoning score: 0.4578125
Action score: 0.4921875
Valid samples: 320/320
Nemotron-RL-agent-workplace_assistant
Dataset Description:
The Nemotron-RL-agent-workplace_assistant is a tool use - multi step agentic environment that tests the agent’s ability to execute tasks in a workplace setting. Workbench contains a sandbox environment with five databases, 26 tools, and 690 tasks. These tasks represent common business activities, such as sending emails, scheduling meetings, etc.
This dataset is released as part of NVIDIA NeMo Gym, a framework for building reinforcement learning environments… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-agent-workplace_assistant.agent-course-final-assignment
Agent Course Final Assignment - Unified Dataset
Author: Arte(r)m Sedov
GitHub: https://github.com/arterm-sedov/
Project link: https://huggingface.co/spaces/arterm-sedov/agent-course-final-assignment
Dataset Description
This dataset is produced by the GAIA Unit 4 Agent for the Hugging Face Agents Course final assignment as part of an experimental multi-LLM agent system that demonstrates advanced AI agent capabilities. It demonstrates advanced AI agent capabilities for… See the full description on the dataset page: https://huggingface.co/datasets/arterm-sedov/agent-course-final-assignment.Nexus-Agents-ToolCalling
Nexus Agents — Tool-Calling Conversations
Synthetic, schema-verified tool-calling conversations for training the Nexus Projects
agents. This is the exact data behind
Nemotron-3-Nano-30B-A3B — Nexus Agents (GGUF),
including the verification transcripts that scored it (27/27 on the behavioral
interview eval, vs 13/27 for the base model).
Links: the fine-tuned model →
Nemotron-3-Nano-30B-A3B — Nexus Agents (GGUF) ·
the generator + seed data + eval harness →
Nexus Training Studio ·… See the full description on the dataset page: https://huggingface.co/datasets/NexusProjectsAI/Nexus-Agents-ToolCalling.Nemotron-RL-Agentic-Function-Calling-Pivot-v1
Dataset Description:
This is a RL dataset for general function-calling by utilizing existing expert tool-use trajectories. We pose each assistant step of the trajectory as a separate behavior cloning problem where the policy model is incentivized to match the tool call choices of the expert model.
This dataset is released as part of NVIDIA NeMo Gym, a framework for building reinforcement learning environments to train large language models. NeMo Gym contains a growing collection of… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Agentic-Function-Calling-Pivot-v1.agentic-ai-options-resultsNemotron-RL-Agentic-Terminal-Pivot-v1
Dataset Description
The Nemotron-RL-Agentic-Terminal-Pivot-v1 dataset provides training samples for reinforcement learning of command-line ("terminal use") LLM agents with the terminus_judge environment in NeMo Gym.
Each record is a single agent decision point extracted from a successful agent trajectory on a terminal task:
responses_create_params.input — the prompt: the task instruction plus the terminal interaction history (prior agent actions and terminal outputs) up to the… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Agentic-Terminal-Pivot-v1.GLM-5.2-AgentThis dataset was generated using teich by TeichAI
GLM-5.2 Agent traces
This directory contains raw agent trace files generated by teich.
JSONL files: 319
Model metadata: glm-5.2
Training-ready tools
Generated agent traces carry configured or recovered tool schemas so tools remain available for training even when a session did not call them.
Native Claude Code imports recover schemas for Claude Code and Claude Desktop built-ins, plus conservative name-derived MCP… See the full description on the dataset page: https://huggingface.co/datasets/AletheiaResearch/GLM-5.2-Agent.TIR-Bench
TIR-Bench: A Comprehensive Benchmark for Agentic Thinking-with-Images Reasoning
Introduction:
TIR-Bench is a comprehensive benchmark designed to evaluate the "thinking-with-images" capabilities of Multimodal Large Language Models (MLLMs), addressing a gap left by existing benchmarks like Visual Search which only test basic operations. As models like OpenAI o3 begin to intelligently create and operate tools to transform images for problem-solving, TIR-Bench provides 13… See the full description on the dataset page: https://huggingface.co/datasets/Agents-X/TIR-Bench.AgentPack
AgentPack
AgentPack is a dataset of 1.3M commits co-authored by Claude Code, OpenAI Codex, and Cursor Agent in public GitHub projects from April up to mid-August 2025.
Dataset Details
See our paper for details.
How to use
Simply load the data using datasets:
import datasets
agentpack_ds = datasets.load_dataset("nuprl/AgentPack", data_dir="train")
Citation
@misc{agentpack,
title={AgentPack: A Dataset of Code Changes… See the full description on the dataset page: https://huggingface.co/datasets/nuprl/AgentPack.Nemotron-RL-Agentic-Conversational-Tool-Use-Pivot-v1
Dataset Description:
We created an RL dataset for conversational tool-use by utilizing existing expert tool-use trajectories. We pose each assistant step of the trajectory as a separate behavior cloning problem where the policy model is incentivized to match the tool call choices of the expert model. Each trajectory includes the use of tools for authentication, data lookup, servicing (i.e. booking reservations, changing them, getting discounts, etc), and more across 838 different… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Agentic-Conversational-Tool-Use-Pivot-v1.Nemotron-RL-Agentic-Indirect-Prompt-Injection-v1
Nemotron-RL-Agentic-Indirect-Prompt-Injection-v1
Dataset Description:
Nemotron-RL-Agentic-Indirect-Prompt-Injection-v1 is an RL dataset for training and evaluating a tool-using agent's ability to resist Indirect Prompt Injection (IPI) attacks hidden inside tool-returned environment data. In each record, the agent receives a benign user request that requires calling a read tool whose output contains an adversarial instruction disguised as legitimate domain content… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Agentic-Indirect-Prompt-Injection-v1.DeepSeek-v4-Pro-AgentThis dataset was generated using teich by TeichAI
Prepare these datasets for supervised fine-tuning in just a few lines of code — see the Conversion section below.
DeepSeek v4 Pro Agent Traces
This directory contains raw agent trace files generated by teich.
All assistant responses were generated by deepseek/deepseek-v4-pro.
JSONL files: 4006
Training-ready tools
A complete configured tools schema snapshot is embedded in the collapsed section at the bottom of… See the full description on the dataset page: https://huggingface.co/datasets/ronaldcmz/DeepSeek-v4-Pro-Agent.AgentWorldBench
AgentWorldBench
AgentWorldBench is a comprehensive evaluation benchmark for language world models, constructed from real-world observations of frontier model trajectories on established benchmarks such as Tool Decathlon, Terminal-Bench 1.0 & 2.0, and OSWorld-Verified. Every evaluation sample is paired with a ground-truth observation obtained from real environment execution, enabling reference-grounded scoring.
AgentWorldBench evaluates world modeling quality by scoring each… See the full description on the dataset page: https://huggingface.co/datasets/Qwen/AgentWorldBench.Nemotron-RL-Agentic-SWE-Pivot-v1
Dataset Description:
The SWE-RL dataset provides GitHub issues for training and validating real-world software engineering agents using the OpenHands environment in NeMo Gym. The dataset is a refactored version of the SWE-Gym and R2E-Gym datasets to support the NeMo Gym input format.
This dataset is released as part of NVIDIA NeMo Gym, a framework for building reinforcement learning environments to train large language models. NeMo Gym contains a growing collection of training… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Agentic-SWE-Pivot-v1.arc-agi3-kimi-k2.7-ar25
ARC-AGI-3 ar25 — Agent Trajectories (kimi-k2.7)
Gameplay trajectories from the harness×model pair kimi-k2.7 playing the
ARC-AGI-3 game ar25, part of the
ARA-as-world-model generalization experiment. The agent builds a structured world model
(an Agent-Native Research Artifact) live during play and consults it to crack levels it
cannot solve from cold exploration.
One dataset repo per harness×model×game: sibling repos
arc-agi3-<harness>-<model>-<game> hold the same game played by… See the full description on the dataset page: https://huggingface.co/datasets/AgentNativeResearchLab/arc-agi3-kimi-k2.7-ar25.gpt-5.5-agentThis dataset was generated using teich by TeichAI
Prepare these datasets for supervised fine-tuning in just a few lines of code — see the Conversion section below.
gpt 5.5 Agent Traces
This directory contains raw agent trace files generated by teich. (I also dropped in some of my own personal traces)
All assistant responses were generated by openai/gpt-5.5.
JSONL files: 88
Training-ready tools
A complete configured tools schema snapshot is embedded in the… See the full description on the dataset page: https://huggingface.co/datasets/armand0e/gpt-5.5-agent.mcp-agent-trajectory-benchmark
MCP Agent Trajectory Benchmark
A benchmark dataset of 49 MCP (Model Context Protocol) agent trajectories (38 single-pass + 11 multi-conv) with complete tool-use traces in the ATIF v1.2 (Agent Trajectory Interchange Format) format. Each agent operates in a distinct business domain with custom tools, realistic user conversations, and full execution traces.
Designed for training and evaluating tool-use / function-calling capabilities of LLMs.
Overview
Item
Details… See the full description on the dataset page: https://huggingface.co/datasets/obaydata/mcp-agent-trajectory-benchmark.multi-agent-coordination-transcripts
Multi Agent Coordination Transcripts
Rights & intended use: legacy public research corpus / portfolio
artifact. Hosted frontier-model outputs are research-only inputs under
project policy (synthetic-factory#161):
intended_use: research_only, project_training_policy: blocked. Not
training data for any model-weight update. Machine-readable record:
rights.json.
Release status: The raw, uncurated payload is now published under
data/raw/. It is available for inspection and… See the full description on the dataset page: https://huggingface.co/datasets/rmems/multi-agent-coordination-transcripts.train-of-thought
Train of Thought Dataset
Overview
This dataset readapts agentlans/think-more
into the Alpaca-style instruction tuning format for training language models in direct answering and chain-of-thought reasoning.
Dataset Structure
Each original example was randomly assigned to be thinking on or off:
Thinking off: Outputs only the final answer.
Thinking on:
Outputs a chain-of-thought (CoT) reasoning process wrapped in <think>...</think>, followed by the final answer… See the full description on the dataset page: https://huggingface.co/datasets/agentlans/train-of-thought.agentabstain
AgentAbstain: Do LLM Agents Know When Not to Act?
AgentAbstain is a paired-task benchmark for agentic abstention: the calibrated ability of tool-using LLM agents to recognize when not to act. It contains 263 task pairs across 42 executable MCP sandbox environments, built on an agent-native taxonomy of 8 abstention scenarios. Every should-act task ships with a should-abstain variant that differs by a single controlled perturbation to the instruction, the… See the full description on the dataset page: https://huggingface.co/datasets/antiquality/agentabstain.AgentSynth
AgentSynth
AgentSynth: Scalable Task Generation for Generalist Computer-Use Agents
Paper | Project Page | Code
Abstract
We introduce AgentSynth, a scalable and cost-efficient pipeline for automatically synthesizing high-quality tasks and trajectory datasets for generalist computer-use agents. Leveraging information asymmetry, AgentSynth constructs subtasks that are simple during generation but significantly more challenging when composed into long-horizon… See the full description on the dataset page: https://huggingface.co/datasets/sunblaze-ucb/AgentSynth.