datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
grounded-visual-spatial-reasoning
Grounded Visual Spatial Reasoning
Code for generating the annotations can be found here: github.com
Dataset Summary
This dataset extends the Visual Spatial Reasoning (VSR) dataset with visual grounding annotations: each caption is annotated with COCO-category object mentions, their positions , and corresponding bounding boxes in the image.
Data instance
Each sample instance has the following structure:
Field
Type
Description
image_file
string… See the full description on the dataset page: https://huggingface.co/datasets/tomhodemon/grounded-visual-spatial-reasoning.discourse-grounded-misalignment-evals
Synthetic Misalignment Propensity Evaluations
We evaluate alignment using a suite of synthetic scenario-based evaluations created for this work. Each question presents
the AI with a high-stakes setting requiring a choice between two actions: one aligned and one misaligned. The misaligned
action is typically framed as instrumentally appealing, making these evaluations a relevant proxy for misaligned AIs across
a range of terminal goals (Bostrom, 2012).
We measure tendencies toward… See the full description on the dataset page: https://huggingface.co/datasets/geodesic-research/discourse-grounded-misalignment-evals.risale-nur-grounded-multipool
Risale-i Nur Grounded Multi-Pool LLM Dataset
TR. 15 kanonik Risale-i Nur kitabından hazırlanan; kaynak
bağlı üretim, SFT, tercih, değerlendirme, sürekli ön eğitim ve erişim
çalışmaları için çok görünümlü bir veri seti.
EN. A multi-view dataset built from 15 canonical Risale-i
Nur books for grounded generation, SFT, preference learning, evaluation,
continued pretraining, and retrieval.
v2.10.0 · 199 configs · 463 config/split views ·
527,196 rows across configured views… See the full description on the dataset page: https://huggingface.co/datasets/risaleinur/risale-nur-grounded-multipool.Grounded-RAG-RU-v2
Датасет для алайнмента (граундинга) способности LLM отвечать на вопросы по документам (RAG)
Этот датасет был собран на основе 13к разных статей из русской Википедии с помошью синтетических вопросов и ответов gpt-4-turbo-1106.
Датасет содержит 4047 уникальных кластеров, т.е. комбинаций из документов - улосвная симуляция "найденных результатов" в Retrieval системе. Подробнее описано в разделе "Общие этапы сборки этого датасета".
Общий объем датасета - 50210 уникальных диалогов.
В… See the full description on the dataset page: https://huggingface.co/datasets/Vikhrmodels/Grounded-RAG-RU-v2.so101_grounded_1fps_test_v1This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 30,
"features": {
"observation.images.top": {
"dtype": "video",
"shape": [
480,
640,
3
],
"names": [
"height",
"width",
"channels"
],
"info": {
"video.height":… See the full description on the dataset page: https://huggingface.co/datasets/kdaterao/so101_grounded_1fps_test_v1.so101_grounded_batched_test_v7This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 30,
"features": {
"observation.images.top": {
"dtype": "video",
"shape": [
480,
640,
3
],
"names": [
"height",
"width",
"channels"
],
"info": {
"video.height":… See the full description on the dataset page: https://huggingface.co/datasets/kdaterao/so101_grounded_batched_test_v7.so101_grounded_batched_test_v6This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 30,
"features": {
"observation.images.top": {
"dtype": "video",
"shape": [
480,
640,
3
],
"names": [
"height",
"width",
"channels"
],
"info": {
"video.height":… See the full description on the dataset page: https://huggingface.co/datasets/kdaterao/so101_grounded_batched_test_v6.so101_grounded_batched_test_v12This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 30,
"features": {
"observation.images.top": {
"dtype": "video",
"shape": [
480,
640,
3
],
"names": [
"height",
"width",
"channels"
],
"info": {
"video.height":… See the full description on the dataset page: https://huggingface.co/datasets/kdaterao/so101_grounded_batched_test_v12.so101_grounded_batched_test_v5This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 30,
"features": {
"observation.images.top": {
"dtype": "video",
"shape": [
480,
640,
3
],
"names": [
"height",
"width",
"channels"
],
"info": {
"video.height":… See the full description on the dataset page: https://huggingface.co/datasets/kdaterao/so101_grounded_batched_test_v5.so101_grounded_batched_test_v10This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 30,
"features": {
"observation.images.top": {
"dtype": "video",
"shape": [
480,
640,
3
],
"names": [
"height",
"width",
"channels"
],
"info": {
"video.height":… See the full description on the dataset page: https://huggingface.co/datasets/kdaterao/so101_grounded_batched_test_v10.so101_grounded_molmo_batch_test_v1This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 30,
"features": {
"observation.images.top": {
"dtype": "video",
"shape": [
480,
640,
3
],
"names": [
"height",
"width",
"channels"
],
"info": {
"video.height":… See the full description on the dataset page: https://huggingface.co/datasets/kdaterao/so101_grounded_molmo_batch_test_v1.so101_grounded_single_testThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 30,
"features": {
"observation.images.top": {
"dtype": "video",
"shape": [
480,
640,
3
],
"names": [
"height",
"width",
"channels"
],
"info": {
"video.height":… See the full description on the dataset page: https://huggingface.co/datasets/kdaterao/so101_grounded_single_test.RAG-Grounded-QA-188k
🎯 RAG Grounded QA 186K
The Anti-Hallucination Dataset
Teach language models to answer from context — or shut up trying.
Built by NovachronoAI — Precision AI for the real world.
Full Dataset (186K) · 20K Subset · Schema · Sources · Usage Guide
🧠 Why This Dataset Exists
Most QA datasets teach models what to say. This one also teaches them when to stay silent.
RAG (Retrieval-Augmented Generation) systems have a fatal flaw: the model hallucinates when… See the full description on the dataset page: https://huggingface.co/datasets/NovachronoAI/RAG-Grounded-QA-188k.grounded-misunderstandings-in-maptask
GMMT: Grounded Misunderstandings in MapTask
The Grounded Misunderstandings in MapTask (GMMT) dataset was produced for the LREC 2026
paper Grounded Misunderstandings in Asymmetric Dialogue: A Perspectivist Annotation
Scheme for MapTask by Nan Li, Albert Gatt, and Massimo Poesio.
It provides perspectivist annotations of the HCRC MapTask corpus,
capturing both speaker-intended and addressee-interpreted landmarks for every
reference expression (RE). The annotations support… See the full description on the dataset page: https://huggingface.co/datasets/chnln/grounded-misunderstandings-in-maptask.Grounded-RAG-QA-RU
Датасет для алайнмента (граундинга) способности LLM отвечать на вопросы по документам (RAG)
Этот датасет был собран на основе 13к статей из русской Википедии с помошью синтетических вопросов и ответов gpt-4-turbo-1106.
В колонке conversation представлен диалог в huggingface формате с ролями: documents, user, assistant. Длина диалогов подразумевает модели с 16k контекста.
Важно: По логике граундинга этого подхода - роль assistant специально повторяется дважды в конце. Первый ответ… See the full description on the dataset page: https://huggingface.co/datasets/Vikhrmodels/Grounded-RAG-QA-RU.lerobot_so101_grounded_sim_pickplaceThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": "so101",
"total_episodes": 105,
"total_frames": 27391,
"total_tasks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:105"
},
"data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet",
"video_path": "videos/{video_key}/chunk-{chunk_index:03d}/file-{file_index:03d}.mp4",
"features":… See the full description on the dataset page: https://huggingface.co/datasets/davidlinjiahao/lerobot_so101_grounded_sim_pickplace.grounded-dataset-final-testThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 30,
"features": {
"observation.images.top": {
"dtype": "video",
"shape": [
480,
640,
3
],
"names": [
"height",
"width",
"channels"
],
"info": {
"video.height":… See the full description on the dataset page: https://huggingface.co/datasets/kdaterao/grounded-dataset-final-test.dfm14-ru-synthetic-grounded-instruct-r1
dfm14-ru-synthetic-grounded-instruct-r1
Accepted-only DFM14 data, automatically reviewed with Gemma 4 26B-A4B. This is not human quality certification. Rejections, repair requests and failed reviews are excluded. Some early synthetic accepts predate the compact reviewer; their audit provenance is retained.
Rows: 9,850. All final assistant targets fit the native Gemma 4 template within 4,096 tokens, without truncation. Complete chat history is preserved. Training supervises the… See the full description on the dataset page: https://huggingface.co/datasets/schneiderkamplab/dfm14-ru-synthetic-grounded-instruct-r1.dfm14-vi-synthetic-grounded-instruct-r1
dfm14-vi-synthetic-grounded-instruct-r1
Accepted-only DFM14 data, automatically reviewed with Gemma 4 26B-A4B. This is not human quality certification. Rejections, repair requests and failed reviews are excluded. Some early synthetic accepts predate the compact reviewer; their audit provenance is retained.
Rows: 9,868. All final assistant targets fit the native Gemma 4 template within 4,096 tokens, without truncation. Complete chat history is preserved. Training supervises the… See the full description on the dataset page: https://huggingface.co/datasets/schneiderkamplab/dfm14-vi-synthetic-grounded-instruct-r1.devim-grounded-turkish
DEVİM Grounded Turkish
This is a methodology and public-evidence repository, not a release of the underlying rights-restricted Turkish corpus.
DEVİM's grounded-data program was created to reduce shortcut learning and weak transfer by linking supervision to source evidence, preserving provenance, separating training material from sequestered evaluation material, and explicitly testing abstention when an answer is not supported.
Verified source frame
Authorized… See the full description on the dataset page: https://huggingface.co/datasets/bazobehram/devim-grounded-turkish.dfm14-ar-synthetic-grounded-instruct-r1
dfm14-ar-synthetic-grounded-instruct-r1
Accepted-only DFM14 data, automatically reviewed with Gemma 4 26B-A4B. This is not human quality certification. Rejections, repair requests and failed reviews are excluded. Some early synthetic accepts predate the compact reviewer; their audit provenance is retained.
Rows: 9,961. All final assistant targets fit the native Gemma 4 template within 4,096 tokens, without truncation. Complete chat history is preserved. Training supervises the… See the full description on the dataset page: https://huggingface.co/datasets/schneiderkamplab/dfm14-ar-synthetic-grounded-instruct-r1.dfm14-ja-synthetic-grounded-instruct-r1
dfm14-ja-synthetic-grounded-instruct-r1
Accepted-only DFM14 data, automatically reviewed with Gemma 4 26B-A4B. This is not human quality certification. Rejections, repair requests and failed reviews are excluded. Some early synthetic accepts predate the compact reviewer; their audit provenance is retained.
Rows: 9,513. All final assistant targets fit the native Gemma 4 template within 4,096 tokens, without truncation. Complete chat history is preserved. Training supervises the… See the full description on the dataset page: https://huggingface.co/datasets/schneiderkamplab/dfm14-ja-synthetic-grounded-instruct-r1.dfm14-ga-synthetic-grounded-instruct-r1
dfm14-ga-synthetic-grounded-instruct-r1
Accepted-only DFM14 data, automatically reviewed with Gemma 4 26B-A4B. This is not human quality certification. Rejections, repair requests and failed reviews are excluded. Some early synthetic accepts predate the compact reviewer; their audit provenance is retained.
Rows: 17,568. All final assistant targets fit the native Gemma 4 template within 4,096 tokens, without truncation. Complete chat history is preserved. Training supervises the… See the full description on the dataset page: https://huggingface.co/datasets/schneiderkamplab/dfm14-ga-synthetic-grounded-instruct-r1.dfm14-mk-synthetic-grounded-instruct-r1
dfm14-mk-synthetic-grounded-instruct-r1
Accepted-only DFM14 data, automatically reviewed with Gemma 4 26B-A4B. This is not human quality certification. Rejections, repair requests and failed reviews are excluded. Some early synthetic accepts predate the compact reviewer; their audit provenance is retained.
Rows: 19,827. All final assistant targets fit the native Gemma 4 template within 4,096 tokens, without truncation. Complete chat history is preserved. Training supervises the… See the full description on the dataset page: https://huggingface.co/datasets/schneiderkamplab/dfm14-mk-synthetic-grounded-instruct-r1.dfm14-tr-synthetic-grounded-instruct-r1
dfm14-tr-synthetic-grounded-instruct-r1
Accepted-only DFM14 data, automatically reviewed with Gemma 4 26B-A4B. This is not human quality certification. Rejections, repair requests and failed reviews are excluded. Some early synthetic accepts predate the compact reviewer; their audit provenance is retained.
Rows: 9,858. All final assistant targets fit the native Gemma 4 template within 4,096 tokens, without truncation. Complete chat history is preserved. Training supervises the… See the full description on the dataset page: https://huggingface.co/datasets/schneiderkamplab/dfm14-tr-synthetic-grounded-instruct-r1.dfm14-cy-synthetic-grounded-instruct-r1
dfm14-cy-synthetic-grounded-instruct-r1
Accepted-only DFM14 data, automatically reviewed with Gemma 4 26B-A4B. This is not human quality certification. Rejections, repair requests and failed reviews are excluded. Some early synthetic accepts predate the compact reviewer; their audit provenance is retained.
Rows: 18,956. All final assistant targets fit the native Gemma 4 template within 4,096 tokens, without truncation. Complete chat history is preserved. Training supervises the… See the full description on the dataset page: https://huggingface.co/datasets/schneiderkamplab/dfm14-cy-synthetic-grounded-instruct-r1.dfm14-gl-synthetic-grounded-instruct-r1
dfm14-gl-synthetic-grounded-instruct-r1
Accepted-only DFM14 data, automatically reviewed with Gemma 4 26B-A4B. This is not human quality certification. Rejections, repair requests and failed reviews are excluded. Some early synthetic accepts predate the compact reviewer; their audit provenance is retained.
Rows: 19,852. All final assistant targets fit the native Gemma 4 template within 4,096 tokens, without truncation. Complete chat history is preserved. Training supervises the… See the full description on the dataset page: https://huggingface.co/datasets/schneiderkamplab/dfm14-gl-synthetic-grounded-instruct-r1.dfm14-mt-synthetic-grounded-instruct-r1
dfm14-mt-synthetic-grounded-instruct-r1
Accepted-only DFM14 data, automatically reviewed with Gemma 4 26B-A4B. This is not human quality certification. Rejections, repair requests and failed reviews are excluded. Some early synthetic accepts predate the compact reviewer; their audit provenance is retained.
Rows: 18,565. All final assistant targets fit the native Gemma 4 template within 4,096 tokens, without truncation. Complete chat history is preserved. Training supervises the… See the full description on the dataset page: https://huggingface.co/datasets/schneiderkamplab/dfm14-mt-synthetic-grounded-instruct-r1.dfm14-id-synthetic-grounded-instruct-r1
dfm14-id-synthetic-grounded-instruct-r1
Accepted-only DFM14 data, automatically reviewed with Gemma 4 26B-A4B. This is not human quality certification. Rejections, repair requests and failed reviews are excluded. Some early synthetic accepts predate the compact reviewer; their audit provenance is retained.
Rows: 9,923. All final assistant targets fit the native Gemma 4 template within 4,096 tokens, without truncation. Complete chat history is preserved. Training supervises the… See the full description on the dataset page: https://huggingface.co/datasets/schneiderkamplab/dfm14-id-synthetic-grounded-instruct-r1.dfm14-hi-synthetic-grounded-instruct-r1
dfm14-hi-synthetic-grounded-instruct-r1
Accepted-only DFM14 data, automatically reviewed with Gemma 4 26B-A4B. This is not human quality certification. Rejections, repair requests and failed reviews are excluded. Some early synthetic accepts predate the compact reviewer; their audit provenance is retained.
Rows: 9,924. All final assistant targets fit the native Gemma 4 template within 4,096 tokens, without truncation. Complete chat history is preserved. Training supervises the… See the full description on the dataset page: https://huggingface.co/datasets/schneiderkamplab/dfm14-hi-synthetic-grounded-instruct-r1.