vpt
Datasets
All datasets matching “vpt”minecraft-vpt-mp4
Minecraft VPT MP4 ArrayRecords
Minecraft gameplay clips and their aligned VPT-style actions, packaged as
sharded ArrayRecord files. This is the
raw-video dataset used by the Minecraft data path in dreamer4-jax-private.
Download
Install the Hugging Face CLI and download the repository to a local directory:
pip install -U huggingface_hub
hf download reactor-team/minecraft-vpt-mp4 \
--repo-type dataset \
--local-dir /path/to/mp4-arrayrecords
The downloaded… See the full description on the dataset page: https://huggingface.co/datasets/reactor-team/minecraft-vpt-mp4.vpt-owamcapThis dataset is an OWAMcap conversion from the Video PreTraining (VPT) minecraft dataset.
It is compressed with the WebDataset Format, which is essentially a series of compressed tar files.
Each sample contains:
.mp4 (containing video, from the original VPT dataset)
.jsonl (containing actions, from the original VPT dataset)
.mcap (OWAMcap format containing actions, which is converted from jsonl)
There are 26322 numbers of samples, a total volume of 5.2TB.
For reading OWAMcap data, please… See the full description on the dataset page: https://huggingface.co/datasets/open-world-agents/vpt-owamcap.vpt_data_8xx_shard0042This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": null,
"total_episodes": 67,
"total_frames": 315759,
"total_tasks": 1,
"total_videos": 67,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 20,
"splits": {
"train": "0:67"
},
"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/BarryFutureman/vpt_data_8xx_shard0042.pan-2-vpt
pan-2 VPT
Cleaned OpenAI VPT contractor demos, packed for Infatoshi/pan-2.
This is not a re-dump of the original VPT Azure blobs, and it is not the same layout as zhwang4ai/OpenAI-Minecraft-Contractor (jsonl-only), p-doom/openai-minecraft-dataset (Grain/ArrayRecord tars), or TESS-Computer/minecraft-vla-stage1 (parquet JPEGs at 640x360 / 5 Hz).
Episodes
1625 validated stems (27 dropped)
Frames
8,155,382 (~113.3 h at 20 Hz)
Image
uint8 [T, 64, 64, 3]
Actions… See the full description on the dataset page: https://huggingface.co/datasets/Infatoshi/pan-2-vpt.vpt_data_8xx_shard0004This dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v2.1",
"robot_type": null,
"total_episodes": 66,
"total_frames": 346970,
"total_tasks": 1,
"total_videos": 66,
"total_chunks": 1,
"chunks_size": 1000,
"fps": 20,
"splits": {
"train": "0:66"
},
"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/BarryFutureman/vpt_data_8xx_shard0004.minecraft-vpt-tokenized
Minecraft VPT Tokenized Latents
Pre-tokenized latent episodes and aligned VPT-style actions derived from the
Minecraft VPT MP4 ArrayRecord dataset. The files are msgpack-encoded
ArrayRecord shards intended for dynamics and world-model training in
dreamer4-jax-private.
Download
pip install -U huggingface_hub
hf download reactor-team/minecraft-vpt-tokenized \
--repo-type dataset \
--local-dir /path/to/tokenized_data_500M
The expected layout is:… See the full description on the dataset page: https://huggingface.co/datasets/reactor-team/minecraft-vpt-tokenized.