xdanielho commited on
Commit
a14fa8b
·
verified ·
1 Parent(s): 9a527a4

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +3 -1
README.md CHANGED
@@ -12,6 +12,8 @@ Download with:
12
  huggingface-cli download 1x-technologies/worldmodel --repo-type dataset --local-dir data
13
  ```
14
 
 
 
15
  Changes from v1.1:
16
  - New train and val dataset of 100 hours, replacing the v1.1 datasets
17
  - Blur applied to faces
@@ -20,7 +22,7 @@ Contents of train/val_v2.0:
20
 
21
  The training dataset is shareded into 100 independent shards. The shapes and definitions of the arrays are as follows (N is the number of frames).
22
 
23
- - **video_{shard}.bin**: 8x8x8 image patches at 30hz, with 17 frame temporal window, encoded using [NVIDIA Cosmos Tokenizer](https://github.com/NVIDIA/Cosmos-Tokenizer) "Cosmos-Tokenizer-DV8x8x8".
24
  - **segment_indicies** - For video `n` and frame `i`, `segment_idx_n[i]` uniquely points to the segment index that frame `i` came from. You may want to use this to separate non-contiguous frames from different videos (transitions).
25
  - **robot_states** - States arrays defined in `Index-to-State Mapping` stored in `np.float32` format. For video `n` and frame `i`, the corresponding state is given by `states_n[i]`.
26
  - **metadata** - The `metadata.json` file provides high-level information about the entire dataset, while `metadata_[n].json` files contain specific details for each individual video `n`.
 
12
  huggingface-cli download 1x-technologies/worldmodel --repo-type dataset --local-dir data
13
  ```
14
 
15
+ **Current version: v2.0**
16
+
17
  Changes from v1.1:
18
  - New train and val dataset of 100 hours, replacing the v1.1 datasets
19
  - Blur applied to faces
 
22
 
23
  The training dataset is shareded into 100 independent shards. The shapes and definitions of the arrays are as follows (N is the number of frames).
24
 
25
+ - **videos**: 8x8x8 image patches at 30hz, with 17 frame temporal window, encoded using [NVIDIA Cosmos Tokenizer](https://github.com/NVIDIA/Cosmos-Tokenizer) "Cosmos-Tokenizer-DV8x8x8".
26
  - **segment_indicies** - For video `n` and frame `i`, `segment_idx_n[i]` uniquely points to the segment index that frame `i` came from. You may want to use this to separate non-contiguous frames from different videos (transitions).
27
  - **robot_states** - States arrays defined in `Index-to-State Mapping` stored in `np.float32` format. For video `n` and frame `i`, the corresponding state is given by `states_n[i]`.
28
  - **metadata** - The `metadata.json` file provides high-level information about the entire dataset, while `metadata_[n].json` files contain specific details for each individual video `n`.