datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
robocasa-mimicgenThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 20,
"features": {
"observation.images.robot0_eye_in_hand": {
"dtype": "video",
"shape": [
256,
256,
3
],
"names": [
"height",
"width",
"channels"
],
"info": {… See the full description on the dataset page: https://huggingface.co/datasets/euijinrnd/robocasa-mimicgen.mimicgen_100_lerobotThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "franka",
"total_episodes": 1200,
"total_frames": 426106,
"total_tasks": 12,
"total_videos": 2400,
"total_chunks": 2,
"chunks_size": 1000,
"fps": 20,
"splits": {
"train": "0:1200"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/locht131/mimicgen_100_lerobot.mimicgen_threading_d0_224x224_mtdit_flow_55k_successThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 20,
"features": {
"action": {
"dtype": "float32",
"shape": [
7
]
},
"observation.state": {
"dtype": "float32",
"shape": [
37
]
},
"observation.sim_state": {
"dtype": "float32",
"shape":… See the full description on the dataset page: https://huggingface.co/datasets/chomeed/mimicgen_threading_d0_224x224_mtdit_flow_55k_success.mimicgen_threading_d0_224x224_mtdit_flow_55k_failureThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 20,
"features": {
"action": {
"dtype": "float32",
"shape": [
7
]
},
"observation.state": {
"dtype": "float32",
"shape": [
37
]
},
"observation.sim_state": {
"dtype": "float32",
"shape":… See the full description on the dataset page: https://huggingface.co/datasets/chomeed/mimicgen_threading_d0_224x224_mtdit_flow_55k_failure.mimicgen_threading_d0_224x224
mimicgen_threading_d0_224x224
MimicGen Threading_D0 demonstrations in LeRobot v3 format, with both cameras rendered at
224x224. Same pipeline and conventions as chomeed/mimicgen_coffee_d1_224x224.
episodes
1000
frames
224508
fps
20
robot
Panda (OSC_POSE, control_freq 20)
cameras
agentview, eye_in_hand -- 224x224, AV1 crf 30
observation.state
37 = 9 proprioceptive + 28 privileged object dims
Provenance
Built from NVIDIA's released… See the full description on the dataset page: https://huggingface.co/datasets/chomeed/mimicgen_threading_d0_224x224.mimicgen_coffee_D0_20This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": null,
"total_episodes": 20,
"total_frames": 4422,
"total_tasks": 1,
"total_videos": 40,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 20,
"splits": {
"train": "0:20"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/YSL2683/mimicgen_coffee_D0_20.mimicgen_1000_lerobotThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "franka",
"total_episodes": 12000,
"total_frames": 4271753,
"total_tasks": 12,
"total_videos": 24000,
"total_chunks": 12,
"chunks_size": 1000,
"fps": 20,
"splits": {
"train": "0:12000"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/locht131/mimicgen_1000_lerobot.mimicgen_hammer_cleanup_d1_224x224_mtdit_flow_35k_failure_seg
MimicGen hammer_cleanup_d1 failures — tail cut at the outcome
Every episode of
chomeed/mimicgen_hammer_cleanup_d1_224x224_mtdit_flow_35k_failure
with its tail cut at the outcome. The approach is kept in full — every clip starts at frame 0.
Same task, same two cameras, same 20 fps, same feature schema.
All 100 episodes kept, 30,343 frames, median clip 240 of 500
source frames (48%).
Task: Open the drawer, place the hammer inside, and close the drawer.
Where the tail… See the full description on the dataset page: https://huggingface.co/datasets/chomeed/mimicgen_hammer_cleanup_d1_224x224_mtdit_flow_35k_failure_seg.mimicgen_square_d0_224x224_mtdit_flow_40k_failure_seg
MimicGen square_d0 failures — tail cut at the outcome
Every episode of
chomeed/mimicgen_square_d0_224x224_mtdit_flow_40k_failure
with its tail cut at the outcome. The approach is kept in full — every clip starts at frame 0.
Same task, same two cameras, same 20 fps, same feature schema.
All 100 episodes kept, 21145 frames, median clip 176 of 400
source frames (44%).
Where the tail is cut
The outcome frame depends on how far the attempt got. All four cases are… See the full description on the dataset page: https://huggingface.co/datasets/chomeed/mimicgen_square_d0_224x224_mtdit_flow_40k_failure_seg.mimicgen_100_stack_three_lerobotThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "franka",
"total_episodes": 100,
"total_frames": 25551,
"total_tasks": 1,
"total_videos": 200,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 20,
"splits": {
"train": "0:100"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/locht131/mimicgen_100_stack_three_lerobot.mimicgen_threading_d0_224x224_insertcam
mimicgen_threading_d0_224x224
MimicGen Threading_D0 demonstrations in LeRobot v3 format, with both cameras rendered at
224x224. Same pipeline and conventions as chomeed/mimicgen_coffee_d1_224x224.
episodes
1000
frames
224508
fps
20
robot
Panda (OSC_POSE, control_freq 20)
cameras
agentview, eye_in_hand -- 224x224, AV1 crf 30
observation.state
37 = 9 proprioceptive + 28 privileged object dims
Provenance
Built from NVIDIA's released… See the full description on the dataset page: https://huggingface.co/datasets/chomeed/mimicgen_threading_d0_224x224_insertcam.mimicgen_threading_d0_224x224_mtdit_flow_55k_success_insertcamThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 20,
"features": {
"action": {
"dtype": "float32",
"shape": [
7
]
},
"observation.state": {
"dtype": "float32",
"shape": [
37
]
},
"observation.sim_state": {
"dtype": "float32",
"shape":… See the full description on the dataset page: https://huggingface.co/datasets/chomeed/mimicgen_threading_d0_224x224_mtdit_flow_55k_success_insertcam.mimicgen_threading_d0_224x224_mtdit_flow_55k_failure_insertcamThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 20,
"features": {
"action": {
"dtype": "float32",
"shape": [
7
]
},
"observation.state": {
"dtype": "float32",
"shape": [
37
]
},
"observation.sim_state": {
"dtype": "float32",
"shape":… See the full description on the dataset page: https://huggingface.co/datasets/chomeed/mimicgen_threading_d0_224x224_mtdit_flow_55k_failure_insertcam.mimicgen_50_lerobotThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "franka",
"total_episodes": 600,
"total_frames": 213468,
"total_tasks": 12,
"total_videos": 1200,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 20,
"splits": {
"train": "0:600"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/locht131/mimicgen_50_lerobot.mimicgen_100_coffee_preparation_lerobotThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "franka",
"total_episodes": 100,
"total_frames": 67868,
"total_tasks": 1,
"total_videos": 200,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 20,
"splits": {
"train": "0:100"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/locht131/mimicgen_100_coffee_preparation_lerobot.mimicgen_threading_d0_224x224_mtdit_flow_55k_success_segThis dataset was created using LeRobot.
Dataset Description
Every episode of chomeed/mimicgen_threading_d0_224x224_mtdit_flow_55k_success cut down to a
single grasp → terminal clip. Nothing else changes: same task, same cameras, same 20 fps, same
feature schema. Approach and post-outcome frames are gone, so what remains is the part of the
episode where the needle is in hand and the insertion is being attempted.
17368 of 22392 source frames survive (78%), across
100 of 100… See the full description on the dataset page: https://huggingface.co/datasets/chomeed/mimicgen_threading_d0_224x224_mtdit_flow_55k_success_seg.mimicgen_threading_d0_224x224_segThis dataset was created using LeRobot.
Dataset Description
Every episode of chomeed/mimicgen_threading_d0_224x224 cut down to a
single grasp → terminal clip. Nothing else changes: same task, same cameras, same 20 fps, same
feature schema. Approach and post-outcome frames are gone, so what remains is the part of the
episode where the needle is in hand and the insertion is being attempted.
170509 of 224508 source frames survive (76%), across
1000 of 1000 episodes.… See the full description on the dataset page: https://huggingface.co/datasets/chomeed/mimicgen_threading_d0_224x224_seg.mimicgen-d0-threadingThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "robomimic",
"total_episodes": 1000,
"total_frames": 224508,
"total_tasks": 1,
"total_videos": 2000,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 20,
"splits": {
"train": "0:1000"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/ankile/mimicgen-d0-threading.mimicgen_threading_d0_224x224_mtdit_flow_55k_failure_segThis dataset was created using LeRobot.
Dataset Description
Every episode of chomeed/mimicgen_threading_d0_224x224_mtdit_flow_55k_failure cut down to a
single grasp → terminal clip. Nothing else changes: same task, same cameras, same 20 fps, same
feature schema. Approach and post-outcome frames are gone, so what remains is the part of the
episode where the needle is in hand and the insertion is being attempted.
15391 of 40000 source frames survive (38%), across
98 of 100… See the full description on the dataset page: https://huggingface.co/datasets/chomeed/mimicgen_threading_d0_224x224_mtdit_flow_55k_failure_seg.mimicgen-square-d0-light-seed42-1000-opaque-rerender-lerobotmimicgen_three_piece_assembly_D0_100This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": null,
"total_episodes": 100,
"total_frames": 33781,
"total_tasks": 1,
"total_videos": 200,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 20,
"splits": {
"train": "0:100"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/YSL2683/mimicgen_three_piece_assembly_D0_100.mimicgen_square_d0_224x224
mimicgen_square_d0_224x224
MimicGen Square_D0 demonstrations in LeRobot v3 format, with both cameras rendered at
224x224. Same pipeline and conventions as chomeed/mimicgen_coffee_d1_224x224. Companion to chomeed/mimicgen_square_d1_224x224: same task with fixed peg placement, and a 14-dim object block versus D1's 17 (D1 adds the peg pose because its peg moves).
episodes
1000
frames
153477
fps
20
robot
Panda (OSC_POSE, control_freq 20)
cameras
agentview… See the full description on the dataset page: https://huggingface.co/datasets/chomeed/mimicgen_square_d0_224x224.mimicgen_hammer_cleanup_d0_224x224
mimicgen_hammer_cleanup_d0_224x224
MimicGen HammerCleanup_D0 demonstrations in LeRobot v3 format, with both cameras rendered at
224x224. Same pipeline and conventions as chomeed/mimicgen_coffee_d1_224x224. Companion to chomeed/mimicgen_hammer_cleanup_d1_224x224: same task, fixed (non-randomised) cabinet placement.
episodes
1000
frames
285359
fps
20
robot
Panda (OSC_POSE, control_freq 20)
cameras
agentview, eye_in_hand -- 224x224, AV1 crf 30… See the full description on the dataset page: https://huggingface.co/datasets/chomeed/mimicgen_hammer_cleanup_d0_224x224.mimicgen_three_piece_assembly_d0_224x224
mimicgen_three_piece_assembly_d0_224x224
MimicGen ThreePieceAssembly_D0 demonstrations in LeRobot v3 format, with both cameras rendered at
224x224. Same pipeline and conventions as chomeed/mimicgen_coffee_d1_224x224. Note this task's demonstrations are not uniformly successful: 18/1000 never reach the full reward, and 298/1000 reach it and then lose it before the final frame, so success genuinely varies within and across episodes.
episodes
1000
frames
336695
fps… See the full description on the dataset page: https://huggingface.co/datasets/chomeed/mimicgen_three_piece_assembly_d0_224x224.mobile_pickup_tubes_mimicgenThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"robot_type": "mobile",
"total_episodes": 20,
"total_frames": 9414,
"total_tasks": 1,
"chunks_size": 1000,
"data_files_size_in_mb": 100,
"video_files_size_in_mb": 500,
"fps": 30,
"splits": {
"train": "0:20"
},
"data_path": "data/chunk-{chunk_index:03d}/file-{file_index:03d}.parquet",
"video_path":… See the full description on the dataset page: https://huggingface.co/datasets/jwhong1209/mobile_pickup_tubes_mimicgen.mimicgen_square_d0_224x224_mtdit_flow_40k_failureThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 20,
"features": {
"action": {
"dtype": "float32",
"shape": [
7
],
"names": null
},
"observation.state": {
"dtype": "float32",
"shape": [
23
],
"names": null
},
"observation.sim_state": {… See the full description on the dataset page: https://huggingface.co/datasets/chomeed/mimicgen_square_d0_224x224_mtdit_flow_40k_failure.mimicgen_square_d1_224x224_mtdit_flow_45k_failureThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 20,
"features": {
"action": {
"dtype": "float32",
"shape": [
7
]
},
"observation.state": {
"dtype": "float32",
"shape": [
26
]
},
"observation.sim_state": {
"dtype": "float32",
"shape":… See the full description on the dataset page: https://huggingface.co/datasets/chomeed/mimicgen_square_d1_224x224_mtdit_flow_45k_failure.mimicgen_hammer_cleanup_d1_224x224_mtdit_flow_35k_failureThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 20,
"features": {
"action": {
"dtype": "float32",
"shape": [
7
]
},
"observation.state": {
"dtype": "float32",
"shape": [
37
]
},
"observation.sim_state": {
"dtype": "float32",
"shape":… See the full description on the dataset page: https://huggingface.co/datasets/chomeed/mimicgen_hammer_cleanup_d1_224x224_mtdit_flow_35k_failure.mimicgen_hammer_cleanup_d1_224x224
mimicgen_hammer_cleanup_d1_224x224
MimicGen HammerCleanup_D1 demonstrations in LeRobot v3 format, with both cameras rendered at
224x224. Same pipeline and conventions as chomeed/mimicgen_coffee_d1_224x224.
episodes
1000
frames
286847
fps
20
robot
Panda (OSC_POSE, control_freq 20)
cameras
agentview, eye_in_hand -- 224x224, AV1 crf 30
observation.state
37 = 9 proprioceptive + 28 privileged object dims
Provenance
Built from NVIDIA's… See the full description on the dataset page: https://huggingface.co/datasets/chomeed/mimicgen_hammer_cleanup_d1_224x224.so101_baseline_plus_mimicgenThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 20,
"features": {
"action": {
"dtype": "float32",
"names": [
"shoulder_pan.pos",
"shoulder_lift.pos",
"elbow_flex.pos",
"wrist_flex.pos",
"wrist_roll.pos",
"gripper.pos"
],
"shape": [
6… See the full description on the dataset page: https://huggingface.co/datasets/hungdo2401/so101_baseline_plus_mimicgen.