datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
rl-run-archive-2026
RL run archive 2026
Archived raw run artifacts (rollout trajectories, rendered frames, policy and optimizer
checkpoints, configs, logs) from simulation reinforcement-learning experiments, published for
long-term preservation and reproducibility.
Layout mirrors the verified backup trees they were copied from:
tilde/20260915-102000/ and taurus/20260915-085631/: batched tar archives. Every archive
carries a per-file SHA-256 manifest inside it; the batch inventories (9998.json.gz… See the full description on the dataset page: https://huggingface.co/datasets/gavinlaw/rl-run-archive-2026.Multi-SWE-RL-Verified
Multi-SWE-RL-Verified
Gold-patch-validated subset of
PrimeIntellect/Multi-SWE-RL-Reupload
(ByteDance's Multi-SWE-RL): 2,232 / 4,703 rows across
C, Go, Java, JavaScript, Rust, and TypeScript that produce a clean reward signal end-to-end.
Default dataset of the multiswe_v1 taskset.
Changes vs upstream
Starting from the 4,703-row re-upload:
C++ dropped wholesale — 0/449 rows passed gold-patch validation in pass 1; the images are
broken for scoring, not merely… See the full description on the dataset page: https://huggingface.co/datasets/PrimeIntellect/Multi-SWE-RL-Verified.hle-gpt-oss-120b-no-python-260222
hle-gpt-oss-120b-no-python-260222
Deep research agent evaluation on rl-rag/hle_text_only (test split).
Results
Metric
Value
pass@4
47.9%
avg@4
26.6%
Trajectory accuracy
26.6% (2292/8632)
Questions
2158
Trajectories
8632 (4 per question)
Avg tool calls
14.5
Full conversations
❌
Model & Setup
Model
gpt-oss-120b
Judge
gpt-4o
Max tool calls
50
Temperature
0.7
Blocked domains
huggingface.co
Tool Usage… See the full description on the dataset page: https://huggingface.co/datasets/rl-rag/hle-gpt-oss-120b-no-python-260222.KodCode-Light-RL-10K
🐱 KodCode: A Diverse, Challenging, and Verifiable Synthetic Dataset for Coding
KodCode is the largest fully-synthetic open-source dataset providing verifiable solutions and tests for coding tasks. It contains 12 distinct subsets spanning various domains (from algorithmic to package-specific knowledge) and difficulty levels (from basic coding exercises to interview and competitive programming challenges). KodCode is designed for both supervised fine-tuning (SFT) and RL tuning.
🕸️… See the full description on the dataset page: https://huggingface.co/datasets/KodCode/KodCode-Light-RL-10K.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
Deep-RL-Course-CertificationINTELLECT-3-RLguru-RL-92k
Revisiting Reinforcement Learning for LLM Reasoning from A Cross-Domain Perspective
Dataset Description
Guru is a curated six-domain dataset for training large language models (LLM) for complex reasoning with reinforcement learning (RL). The dataset contains 91.9K high-quality samples spanning six diverse reasoning-intensive domains, processed through a comprehensive five-stage curation pipeline to ensure both domain diversity and reward verifiability.… See the full description on the dataset page: https://huggingface.co/datasets/IFM/guru-RL-92k.guru-RL-92k-extra-info-compressed
Revisiting Reinforcement Learning for LLM Reasoning from A Cross-Domain Perspective
Note for this extra-info-compressed data version!
The dataset provided in this repository is specifically intended for use with the latest release of VeRL (v0.4.0). Since VeRL rl_dataset.py processes datasets as datasets.Dataset, it is essential that the structure of all Parquet files remains fully consistent. This repository is designed to meet that requirement.
In this repo, the… See the full description on the dataset page: https://huggingface.co/datasets/IFM/guru-RL-92k-extra-info-compressed.Qwen-RLbrowsecomp-gpt-oss-120b-260222
browsecomp-gpt-oss-120b-260222
Deep research agent evaluation on data/browsecomp.jsonl (normal split).
Results
Metric
Value
pass@4
46.8%
avg@4
23.9%
Trajectory accuracy
23.9% (1211/5064)
Questions
1266
Trajectories
5064 (4 per question)
Avg tool calls
26.1
Full conversations
✅
Model & Setup
Model
gpt-oss-120b
Judge
gpt-4o
Max tool calls
50
Temperature
0.7
Blocked domains
huggingface.co
Tool Usage… See the full description on the dataset page: https://huggingface.co/datasets/rl-rag/browsecomp-gpt-oss-120b-260222.LiteCoder-Terminal-RL-preview
LiteCoder-Terminal-RL-preview
Paper | Code | Blog Post
This dataset contains 602 standardized Harbor terminal environments and was released as part of the paper LiteCoder-Terminal: Scaling Long-Horizon Terminal Environments for Learning Language Agents.
Unlike static text-only instructions, these environments are fully executable and are designed to support the training of terminal-based agents.
Environment Generation Pipeline
The lack of high-quality, executable… See the full description on the dataset page: https://huggingface.co/datasets/Lite-Coder/LiteCoder-Terminal-RL-preview.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-Ultra-Training-Blends
Dataset Description:
This dataset provides Reinforcement Learning (RL) and Multi-teacher On-Policy Distillation (MOPD) training-data blends used by the public Nemotron-3-Ultra post-training recipe. The blends are consumed by the NeMo RL training recipes through the NeMo Gym agent framework, in which each prompt is paired with an agent/environment that returns a verifiable or judge-based reward. Each subset is a separate blend; see the recipe for how the blends are used.
The… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-RL-Ultra-Training-Blends.browsecomp-no-scroll-gpt-oss-120b
browsecomp-no-scroll-gpt-oss-120b
Deep research agent evaluation on data/browsecomp.jsonl (normal split).
Results
Metric
Value
pass@4
46.0%
avg@4
22.9%
Trajectory accuracy
22.9% (1160/5064)
Questions
1266
Trajectories
5064 (4 per question)
Avg tool calls
27.0
Full conversations
❌
Model & Setup
Model
gpt-oss-120b
Judge
gpt-4o
Max tool calls
50
Temperature
0.7
Blocked domains
huggingface.co
Tool… See the full description on the dataset page: https://huggingface.co/datasets/rl-rag/browsecomp-no-scroll-gpt-oss-120b.Math-RL-Tasks
Ulam AI Math RL Tasks
Forty original, verifier-backed mathematical reasoning tasks packaged as ten
independent RL environments. The collection spans advanced graduate exercises,
research-style exact computation and structural generalization problems in
algebraic geometry, arithmetic geometry, combinatorics, topology, probability
and spectral analysis.
Each suite pairs a runnable rl_env/ with a preserved blind_run/ by
GPT-5.6 Sol Pro. The model name describes the evaluation actor… See the full description on the dataset page: https://huggingface.co/datasets/ulamai/Math-RL-Tasks.alignment-faking-rl
Transcripts from Towards training-time mitigations for alignment faking in RL
This dataset contains the full evaluation transcripts through the RL runs for all model organisms in our blog post, Towards training-time mitigations for alignment faking in RL.
Each file in encrypted_transcripts/ corresponds to one RL training run.
Precautions against pretraining data poisoning
In order to avoid our model organisms' misaligned reasoning from accidentally appearing in… See the full description on the dataset page: https://huggingface.co/datasets/Anthropic/alignment-faking-rl.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.browsecomp-high-effort-gpt-oss-120b
browsecomp-high-effort-gpt-oss-120b
Deep research agent evaluation on data/browsecomp.jsonl (normal split).
Results
Metric
Value
pass@4
44.1%
avg@4
22.9%
Trajectory accuracy
22.9% (1158/5064)
Questions
1266
Trajectories
5064 (4 per question)
Avg tool calls
55.4
Full conversations
❌
Model & Setup
Model
gpt-oss-120b
Judge
gpt-4o
Max tool calls
100
Temperature
0.7
Blocked domains
huggingface.co
Tool… See the full description on the dataset page: https://huggingface.co/datasets/rl-rag/browsecomp-high-effort-gpt-oss-120b.VR-X-SFT-RL
VR-X: Visual Reasoning Benchmark for UniVR
VR-X contains three independent data blocks:
SFT data organized by capability.
VR-X-RL data for visual-reasoning reinforcement learning.
VR-X-Eval held-out evaluation data.
VR-X-RL and VR-X-Eval are independent from SFT and must be loaded separately.
Public repository paths use anonymous source codes; no source-to-code mapping is
published.
Repository layout
.
├── Robot Manipulation/ # SFT only
│ └── RM-###/
│… See the full description on the dataset page: https://huggingface.co/datasets/ByteDance/VR-X-SFT-RL.rlbenchfail_train_dataset
Guardian: RLBench-Fail Dataset
This dataset is part of the Guardian project: Detecting Robotic Planning and Execution Errors with Vision-Language Models. It contains annotated robotic manipulation failure data generated in the RLBench simulator for training and evaluating Vision-Language Models (VLMs) on failure detection tasks.
Failures are produced by an automated pipeline that procedurally perturbs successful scripted trajectories in simulation, generating diverse planning… See the full description on the dataset page: https://huggingface.co/datasets/paulpacaud/rlbenchfail_train_dataset.Nemotron-Cascade-2-RL-data
Dataset Description:
The Nemotron-Cascade-2-RL dataset is a curated reinforcement learning (RL) dataset blend used to train Nemotron-Cascade-2-30B-A3B model. It includes instruction-following RL, multi-domain RL, on-policy distillation, and software engineering RL (SWE-RL) data.
This dataset is ready for commercial use.
The dataset contains the following subset:
IF-RL
Contains 45,879 training samples for instruction-following RL. Our curation process mainly… See the full description on the dataset page: https://huggingface.co/datasets/nvidia/Nemotron-Cascade-2-RL-data.r2e-dockers-rllm-v1HUI360
HUI360
HUI360: A 360° Egocentric Dataset and Baselines for Human-Robot Interaction Anticipation (IEEE FG 2026)
Open-access skeleton annotations for HUI360, a large-scale 360° egocentric dataset for human-robot interaction anticipation in the wild. This repository provides the annotations as tabular CSV files (one row per detection), ready for training and evaluation with HUI360-Baselines.
Related resources
Resource
Link
Project… See the full description on the dataset page: https://huggingface.co/datasets/rlorlou/HUI360.agent-task-litecoder-terminal-rl-preview
Apptainer pool for hamishivi/agent-task-litecoder-terminal-rl-preview
This repository hosts tmax-compatible SIF images and a unified download manifest. Training data and task archives are in hamishivi/agent-task-litecoder-terminal-rl-preview. The manifest includes earlier images hosted under hamishivi and new images hosted under TMaxxx; the downloader selects the correct repository and immutable commit for each image.
Apptainer images
The pool currently contains… See the full description on the dataset page: https://huggingface.co/datasets/TMaxxx/agent-task-litecoder-terminal-rl-preview.harbor-swesmith-rl-artifacts
Harbor SWE-Smith 强化学习数据产物
本数据集是 Harbor Qwen 工具调用代码智能体强化学习项目使用的冻结任务集，服务于 GRPO、原生价值模型/GAE PPO、训练过程诊断和统一协议评测。
项目已于 2026 年 8 月 30 日完成 P0 评测并进入阶段性归档。本数据集用于保留实验所依赖的数据切分、任务执行文件和审计信息，不代表新的通用代码能力基准。
数据概况
切分
任务数
训练集
187
验证集
42
测试集
38
合计
267
数据覆盖 89 个上游代码仓库。三个切分之间同时执行任务标识和仓库级隔离检查。
正式数据集名称：
swesmith-curated-grpo-267-v1
冻结切分的语义摘要：
ae5df9a3f4a3fc8af44fac420b36529e283839e1bd3de9daba65d5bcda51447d
该值来自 split-manifest.json 的 sha256 字段，用于标识切分语义，不等同于该文件本身的字节级… See the full description on the dataset page: https://huggingface.co/datasets/keryszhan/harbor-swesmith-rl-artifacts.SPADE-Environment-Pool-GPT5.5-ToolUse
SPARE GPT-5.5 Multi-Turn Tool-Use Games v1
A public static pool of 11,039 validated multi-turn tool-use environments generated by GPT-5.5 for SPARE actor training.
Training alignment
Source recipe: Qwen3-30B-A3B 0624 tool-use GAMES configuration
400 rollouts x 24 games/rollout = 9,600 no-reuse games required
11,039 validated games provide 1,439 games of headroom
Six balanced skills: API orchestration, data retrieval, state modification, error recovery, tool… See the full description on the dataset page: https://huggingface.co/datasets/spade-rl/SPADE-Environment-Pool-GPT5.5-ToolUse.Dolci-Think-RL-7B
Dolci-Think-RL-7B
Dataset Summary
Dolci-Think-RL-7B is the reinforcement learning dataset used to train the Olmo-3-7B-Think model.It contains 102,014 prompts designed to elicit deep reasoning across:
Math
Coding
Precise Instruction Following
General Chat
It blends high-quality curated sources with filtering designed for deliberate reasoning.
Dataset Composition
Total Samples: 102,014
Original Dataset Contribution… See the full description on the dataset page: https://huggingface.co/datasets/allenai/Dolci-Think-RL-7B.browsecomp-qwen35-35b-a3b-think
browsecomp-qwen35-35b-a3b-think
Deep research agent evaluation on data/browsecomp.jsonl (normal split).
Results
Metric
Value
pass@4
43.0%
avg@4
24.8%
Trajectory accuracy
24.8% (1258/5064)
Questions
1266
Trajectories
5064 (4 per question)
Avg tool calls
41.1
Full conversations
❌
Model & Setup
Model
Qwen3.5-35B-A3B
Judge
gpt-4o
Max tool calls
50
Temperature
0.7
Blocked domains
huggingface.co
Tool… See the full description on the dataset page: https://huggingface.co/datasets/rl-rag/browsecomp-qwen35-35b-a3b-think.