|
|
--- |
|
|
license: mit |
|
|
--- |
|
|
|
|
|
# ARC-AGI-2 Human testing data |
|
|
|
|
|
This file contains data from human testing sessions on ARC-AGI tasks. |
|
|
|
|
|
Each row represents a single test attempt by a human participant on a specific task-test pair in the "Public Train" or "Public Eval" ARC-AGI-2 datasets. Not all tasks in the released "Public Train" |
|
|
sets were tested, so these results are not comprehensive. This data does not include tasks from "Semi Private Evaluation" or "Private Evaluation" |
|
|
|
|
|
### Column Descriptions |
|
|
|
|
|
- **task_ID**: A unique identifier for the ARC task that was attempted. |
|
|
- **task_set**: Indicates whether the task belongs to the "Public Train" or "Public Eval" dataset. |
|
|
- **test_index**: The index of the test example within the task (typically 0 or 1). |
|
|
- **session_ID**: A unique identifier for the testing session in which this attempt was made. |
|
|
- **start_time_seconds**: The time (in seconds) from the start of the session when this task attempt was begun. |
|
|
- **duration_seconds**: The total time (in seconds) spent on this task attempt. |
|
|
- **submissions**: The number of solution submissions made during this attempt. |
|
|
- **correct_submissions**: The number of correct solution submissions made during this attempt (0 if the participant did not solve the task correctly, 1 or more if they did). |
|
|
|
|
|
For more information about ARC Prize and ARC-AGI-2, see the [technical paper](https://arcprize.org/blog/arc-agi-2-technical-report). |
