Datasetpublic

Aether258/black_smash_05_eventvla

Black Smash 05 EventVLA Keyframes 这是由本地 black_smash_05 转换得到的 EventVLA/LeRobot 格式数据集，并已写入关键帧标注。数据概况 episodes: 232 frames: 175281 fps: 30 robot_type: bimanual_cartesian_delta_20d videos: 464 video views: observation.images.left_wrist, observation.images.right_wrist action_dim: 20 state_dim: 20 关键帧标注每个 episode 的 meta/episodes.jsonl 中包含： keyframe_steps inspect_keyframe_steps eventvla_keyframe_source 当前关键帧来自 state-based weak teacher 导出的六类阶段点：… See the full description on the dataset page: https://huggingface.co/datasets/Aether258/black_smash_05_eventvla.

sourceHugging Faceotherupdated 3mo agoView on Hugging Face

0likes161downloads

Dataset Card

Black Smash 05 EventVLA Keyframes

这是由本地 black_smash_05 转换得到的 EventVLA/LeRobot 格式数据集，并已写入关键帧标注。

数据概况

—episodes: 232
—frames: 175281
—fps: 30
—robottype: `bimanualcartesiandelta20d`
—videos: 464
—video views: observation.images.left_wrist, observation.images.right_wrist
—action_dim: 20
—state_dim: 20

关键帧标注

每个 episode 的 meta/episodes.jsonl 中包含：

—keyframe_steps
—inspect_keyframe_steps
—eventvla_keyframe_source

当前关键帧来自 state-based weak teacher 导出的六类阶段点：

—grasp_tube
—start_pour
—release_tube
—grasp_pestle
—start_grind
—lift_pestle

导出摘要见 meta/eventvla_keyframe_export_summary.json。

目录结构

text

data/chunk-000/*.parquet
videos/chunk-000/observation.images.left_wrist/*.mp4
videos/chunk-000/observation.images.right_wrist/*.mp4
meta/info.json
meta/modality.json
meta/episodes.jsonl
meta/tasks.jsonl
meta/stats_gr00t.json
meta/eventvla_keyframe_export_summary.json

用途

该数据集用于 EventVLA 关键帧记忆训练，可作为 pure_image_keyframe_memory 流程的训练数据。