datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
SEED-Data-Edit-Part1-Openimages
SEED-Data-Edit
SEED-Data-Edit is a hybrid dataset for instruction-guided image editing with a total of 3.7 image editing pairs, which comprises three distinct types of data:
Part-1: Large-scale high-quality editing data produced by automated pipelines (3.5M editing pairs).
Part-2: Real-world scenario data collected from the internet (52K editing pairs).
Part-3: High-precision multi-turn editing data annotated by humans (95K editing pairs, 21K multi-turn rounds with a maximum of 5… See the full description on the dataset page: https://huggingface.co/datasets/AILab-CVC/SEED-Data-Edit-Part1-Openimages.agent-apprenticeship-seed-dataset
Agent Apprenticeship Seed Dataset
The living ecosystem where AI agents run automated workflow loops on any task, improve through execution, and turn each run into reusable work experience + data to improve future agents.
As agents move into long-horizon, economically valuable work, Agent Apprenticeship creates the open infrastructure where real-world tasks generate reusable learning signals and complex workflows advance through agent loops that turn execution into shared… See the full description on the dataset page: https://huggingface.co/datasets/rayrren/agent-apprenticeship-seed-dataset.agent-apprenticeship-seed-dataset_v0.2
Agent Apprenticeship Seed Dataset v0.2
Real-world agent work experience, looped into collective learning.
The living ecosystem where AI agents complete tasks through workflow loops, improve through iterative execution, are evaluated by mentor agents or humans in the loop, and turn completed work into reusable work experience and data to improve future agents.
As agents move into long-horizon, economically valuable work, Agent Apprenticeship creates the open infrastructure where… See the full description on the dataset page: https://huggingface.co/datasets/rayrren/agent-apprenticeship-seed-dataset_v0.2.opensec-seeds
OpenSec Seeds: Incident Response Scenarios for Agent Calibration
This dataset provides 220 taxonomy-stratified security incident scenarios for training and evaluating AI agents on incident response (IR) tasks. Each scenario includes entity definitions, attack kill chains, ground truth labels, and prompt injection payloads designed to test agent calibration under adversarial evidence.
Paper: OpenSec: Measuring Incident Response Agent Calibration Under Adversarial Evidence… See the full description on the dataset page: https://huggingface.co/datasets/Jarrodbarnes/opensec-seeds.mimicgen-square-d0-light-seed42-1000-opaque-rerender-lerobotseedance_general_all_dance_scm_latent_lmdb
Seedance General-All + Dance SCM Latent LMDB
This dataset stores precomputed SCM latents used for TurboT2AV training.
Source mapping: seedance_general_all_dance_mapping.csv
Successful latent samples: 44,305
Shards: 8 LMDB shards under scm_latent_lmdb/shard_00000 ... shard_00007
Video latent shape per sample: (1, 16, 128, 16, 24)
Audio latent shape per sample: (1, 127, 128)
The source mapping combines Seedance general-all data with a dance subset. The mapping contains 44,504… See the full description on the dataset page: https://huggingface.co/datasets/luyu1021/seedance_general_all_dance_scm_latent_lmdb.spring-seed-catalog
Spring Seed Catalog
Cleared seed packets from the community garden intake.
Retained seed entries: 7
Featured seed: SEED-106 — lettuce / Green Wave
Harvest-year span: 2022-2025
Organic entries: 4
Crops (bean/carrot/lettuce/tomato): 2/3/1/1
AInsteinBench
AInsteinBench
AInsteinBench is a benchmark for evaluating the capabilities of AI agents in solving scientific computing problems. It currently supports Einstein Toolkit and Multi-SWE-bench formats of coding questions.
📊 Dataset Overview
AInsteinBench provides 244 scientific computing tasks derived from multiple scientific repositories. These tasks have been verified on execution and also reviewed by corresponding domain experts to verify both software engineering and… See the full description on the dataset page: https://huggingface.co/datasets/ByteDance-Seed/AInsteinBench.mimicgen-square-d0-bg25-seed42-1000-opaque-rerender-lerobotterminal-bench-4-qwen3-8b-embeddings-train20-seed42-successes
Terminal trajectory embedding task subset
20 selected tasks, 1,281 trajectories,
and 123,849 state/action pairs. Only trajectories with numeric reward == 1 are retained.
The saved vectors are exact selected rows of the existing embeddings; the encoder
was not rerun. train/metadata.json matches both tensor row orders.
source_row_indices.json records the original row indices; selection.json
records selection parameters, source checksums and output checksums.
selected_tasks.json… See the full description on the dataset page: https://huggingface.co/datasets/Hkang/terminal-bench-4-qwen3-8b-embeddings-train20-seed42-successes.levir-yolov8n-p2-kvca-selectivity-e2e-seed42forsy-agent-work-trace-seed
Forsy Agent Work Trace Seed v0.1.0
A seed collection of structured agent work traces generated with Forsy Trace Skill.
Forsy Trace Skill is an open skill for capturing AI agent workflows as structured, annotated trajectory data: task context, step traces, tool use, observations, feedback, failures, retries, artifacts, outcomes, and other learning signals.
This dataset mirrors the machine-readable trace exports from the public GitHub repository:… See the full description on the dataset page: https://huggingface.co/datasets/rayrren/forsy-agent-work-trace-seed.er_cost_marginrl_r1_distill_1.5b_compression_n16_b512_32k_lr1e-6_kl0_seed42-rollouts
er_cost_marginrl_r1_distill_1.5b_compression_n16_b512_32k_lr1e-6_kl0_seed42 rollouts
This dataset contains one compressed JSONL shard for every completed training
step. The step and rollout_index columns uniquely locate a rollout within
this training run. Run metadata and per-step row counts are recorded in
rollout_manifest.json.
tinyperson-yolov8n-p2p3p4-oacp-fixedsplit42-0a2ca54-seed43tinyperson-yolov8n-p2p3p4-oacp-fixedsplit42-0a2ca54-seed42levir-yolov8n-p2-kvca-surgical-placement-probe-seed42levir-yolov9t-no-oacp-no-mosaic-seed42tinyperson-yolov8n-p2p3p4-oacp-fixedsplit42-0a2ca54-seed44levir-yolov9t-oacp-no-mosaic-seed42levir-yolov8n-p2p3p4-samc-oacp-current-seed42levir-yolov8n-p2p3p4-samc-seed423D_STU_test_seed2025
3D STU test — point OOD scores (seed 2025)
Bundle: 3D_STU_test_seed2025.tar.gz
Method: NDP+EE reproduce (Mask4Former3D, seed 2025, last-epoch.ckpt)
Val FPR@95 (development): 0.39% on STU val
Frames: 14608 / 51 sequences; scores fmt=%.6f in tar
See 3D_STU_test_seed2025.manifest.json for SHA256 and provenance.
tinyperson-yolov8n-p2p3p4-samc-cp3-no-oacp-seed42levir-yolov9t-oacp-official-e38ce07-retry1-seed42tinyperson-yolov8n-p2p3p4-samc-seed42rlm-trajectories-seed
HotCopy RLM Trajectories (Seed)
A 12-row seed corpus of synthetic Recursive Language Model trajectories
emitted by the HotCopy two-tier agentic CLI (the orchestrator root, sub-call workers).
Why this dataset exists
The Recursive Language Model paper (Zhang, Kraska, Khattab — MIT CSAIL, 2026,
arxiv.org/abs/2512.24601) reports that
"Fine-tuning Qwen3-8B on 1,000 RLM trajectories improved performance 28.3%"
— a strong signal that the shape of RLM execution can be taught from… See the full description on the dataset page: https://huggingface.co/datasets/HotCopyAI/rlm-trajectories-seed.synthetic_demographics_seed
Synthetic Demographic Seeds v1
This is a dataset of 3,541,040 roughly demographically correct demographic seeds and somewhat demographically accurate names all generated from publicly available datasets.
(note there were tradeoffs made with accuracy and what I could tie together, v2 will be more accurate)
get_synthetic_demographics.py contains a method for quickly and randomly selecting batches of demographic seeds.
There is no filtering on this at the moment.
Format… See the full description on the dataset page: https://huggingface.co/datasets/sacrificialpancakes/synthetic_demographics_seed.levir-yolov8n-p2-tophat-gap-ftal-legacy-seed42dd-seedslevir-yolov8n-p2-tophat-plain-ftal-noggcf-seed42