szlgallen/RnG_training_data
RnG training data objaverse_all287k_renders 287,562 Objaverse objects, 40 views each: RGB .webp + depth _depth.png + transforms.json (camera). Total 1,438,645,422,080 bytes (1.31 TiB). shards/shard_00000.tar … shard_00718.tar — 719 shards, ~2 GB each, 400 per-object tars per shard (last one 362) manifest.tsv — shard → member tar → size → shard sha256 objaverse_all287k.txt — the 287,562 object ids (32-hex). Member tar names are the object's storage key: 64-hex… See the full description on the dataset page: https://huggingface.co/datasets/szlgallen/RnG_training_data.
RnG training data
objaverseall287krenders
287,562 Objaverse objects, 40 views each: RGB .webp + depth _depth.png + transforms.json (camera). Total 1,438,645,422,080 bytes (1.31 TiB).
shards/shard_00000.tar…shard_00718.tar— 719 shards, ~2 GB each, 400 per-object tars per shard (last one 362)manifest.tsv— shard → member tar → size → shard sha256objaverse_all287k.txt— the 287,562 object ids (32-hex). Member tar names are the object's storage key: 64-hex sha256 for 179,216 objects, 32-hex object id for 108,339, Sketchfab uid for 7
Restore: tar -xf shard_XXXXX.tar gives back the original per-object tars; put them in one folder.