r2v
Datasets
All datasets matching “r2v”Omni-R2V
Omni-R2V Dataset
Large-scale training data for omni reference-to-video generation
Overview · Task coverage · Quick start · Data format · Citation
From individual reference factors to multi-content and cross-aspect combinations.
Overview
For evaluation, download the companion OmniVBench benchmark, which provides generation instructions and reference media for 813 evaluation cases.
Omni-R2V provides 339,570 processed training samples across seven reference… See the full description on the dataset page: https://huggingface.co/datasets/wxli318/Omni-R2V.r2v-0826-0827-h3-duration-prompt-fix
R2V-0826/0827 补跑批 — duration + prompt 修正版 (14 条)
生成模型: MiniMax-H3 (sglang diffusion server, ref2va, TP4 x Ulysses2, 8xH100)
源清单: R2V-0826_0827-subset_duration_prompt_fix.json (14 条)
PE: H3-Context-IR 提示词增强（prompt 取自源 JSON 的 prompt 字段，源数据无 user_prompt）
编码: MP4 (CRF 14, output_compression 72) + FFV1/FLAC 无损母版 MKV
参考素材: 14 条全部有参考图/视频；其中 5 条含参考视频（每条 1 个），无参考音频
目录结构
<task_id>/<sample_id>.mp4 成片
<task_id>/<sample_id>_lossless.mkv 无损母版
metadata.jsonl… See the full description on the dataset page: https://huggingface.co/datasets/Atriclove/r2v-0826-0827-h3-duration-prompt-fix.R2_VLA_6_items_sampleThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": "rainbow-prealpha",
"total_episodes": 12,
"total_frames": 4829,
"total_tasks": 12,
"total_videos": 0,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 30,
"splits": {
"train": "0:12"
},
"data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
"video_path": null,
"features": {… See the full description on the dataset page: https://huggingface.co/datasets/HumanoidTeam/R2_VLA_6_items_sample.r2vulr2v_baseline_resultsr2vul-curated