Datasets:
Update README.md
Browse files
README.md
CHANGED
|
@@ -17,7 +17,7 @@ dataset_info:
|
|
| 17 |
struct:
|
| 18 |
- name: x
|
| 19 |
dtype: float32
|
| 20 |
-
- name: y
|
| 21 |
dtype: float32
|
| 22 |
- name: width
|
| 23 |
dtype: float32
|
|
@@ -39,13 +39,22 @@ dataset_info:
|
|
| 39 |
sequence: string
|
| 40 |
splits:
|
| 41 |
- name: train
|
| 42 |
-
num_bytes: 246119840
|
| 43 |
num_examples: 350
|
| 44 |
download_size: 221186675
|
| 45 |
-
dataset_size: 246119840
|
| 46 |
configs:
|
| 47 |
- config_name: default
|
| 48 |
data_files:
|
| 49 |
- split: train
|
| 50 |
path: data/train-*
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 51 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17 |
struct:
|
| 18 |
- name: x
|
| 19 |
dtype: float32
|
| 20 |
+
- name: 'y'
|
| 21 |
dtype: float32
|
| 22 |
- name: width
|
| 23 |
dtype: float32
|
|
|
|
| 39 |
sequence: string
|
| 40 |
splits:
|
| 41 |
- name: train
|
| 42 |
+
num_bytes: 246119840
|
| 43 |
num_examples: 350
|
| 44 |
download_size: 221186675
|
| 45 |
+
dataset_size: 246119840
|
| 46 |
configs:
|
| 47 |
- config_name: default
|
| 48 |
data_files:
|
| 49 |
- split: train
|
| 50 |
path: data/train-*
|
| 51 |
+
license: mit
|
| 52 |
+
task_categories:
|
| 53 |
+
- visual-question-answering
|
| 54 |
+
language:
|
| 55 |
+
- en
|
| 56 |
---
|
| 57 |
+
|
| 58 |
+
# IDM-Single
|
| 59 |
+
|
| 60 |
+
IDM-Single is part of the IDMBench in AgentStudio for evaluating agent's ability to learn from Internet-scale videos without action labels. Specifically, this dataset requires VLMs to answer the action transition between two screenshots. Please note that this dataset is a test set rather than a training set. Therefore, please do not use it for training. More details are provided in the [project page](https://computer-agents.github.io/agent-studio).
|