changelog.md

@@ -4,15 +4,9 @@ This document tracks changes and updates to the ARC-AGI-2 dataset tasks.
 
 ## Updates
 
-### 2025-04-17
-
-* Public eval task `d8e07eb2` - [Single train pair update](https://github.com/arcprize/ARC-AGI-2/commit/14fba87526c727b80b3a9b85d5933fd7825b991f)
-
 ### 2025-04-14
 
 * Public Eval Tasks were updated with minor adjustments (off-by-one-pixel errors and slight ambiguities) to train and test pairs. No major task refactors. Updated tasks:
-    * `38007db0` - [Single test pair update](https://github.com/arcprize/ARC-AGI-2/commit/385b761253cf7157ad503909f4d8224b8d85ca97#diff-41216bd1be9cb219575a44e2a21a7dcf18667c75dfa292d52ea7878a3148bcd1)
-    * `36a08778` - [Single test pair update](https://github.com/arcprize/ARC-AGI-2/commit/385b761253cf7157ad503909f4d8224b8d85ca97)
     * `247ef758` - [Single test pair update](https://github.com/arcprize/ARC-AGI-2/commit/8b454b595552981fc9aa8e9540f3e68c92b0f03a)
     * `f560132c` - [Single test pair update](https://github.com/arcprize/ARC-AGI-2/commit/30c145f7c524c932d95d4a512abdd5318ef21bf9)
     * `f931b4a8` - [Train pair update](https://github.com/arcprize/ARC-AGI-2/commit/86a8149f53ce915c069cf586f061eb0af0204713)

data/evaluation/b6f77b65.json

@@ -152,6 +152,36 @@
       }
     ],
     "test": [
+      {
+        "input": [
+          [3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
+          [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
+          [0, 0, 4, 3, 3, 0, 0, 0, 0, 0, 0, 0],
+          [0, 0, 4, 0, 3, 0, 8, 7, 7, 7, 0, 0],
+          [0, 0, 4, 0, 3, 0, 8, 0, 0, 7, 0, 0],
+          [0, 0, 4, 0, 3, 0, 8, 0, 0, 7, 0, 0],
+          [0, 0, 6, 5, 5, 5, 5, 5, 0, 7, 0, 0],
+          [0, 0, 6, 0, 0, 0, 0, 5, 0, 7, 0, 0],
+          [0, 0, 6, 0, 0, 0, 0, 5, 0, 7, 0, 0],
+          [0, 3, 1, 1, 1, 0, 2, 2, 2, 2, 9, 0],
+          [0, 3, 0, 0, 1, 0, 2, 0, 0, 0, 9, 0],
+          [0, 3, 0, 0, 1, 0, 2, 0, 0, 0, 9, 0]
+        ],
+        "output": [
+          [3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
+          [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
+          [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
+          [0, 0, 0, 0, 0, 0, 0, 0, 0, 7, 0, 0],
+          [0, 0, 4, 0, 0, 0, 0, 0, 0, 7, 0, 0],
+          [0, 0, 4, 0, 0, 0, 8, 7, 7, 7, 0, 0],
+          [0, 0, 4, 0, 0, 0, 8, 5, 0, 7, 0, 0],
+          [0, 0, 4, 0, 0, 0, 8, 5, 0, 7, 0, 0],
+          [0, 0, 6, 5, 5, 5, 5, 5, 0, 7, 0, 0],
+          [0, 0, 6, 0, 1, 0, 2, 2, 2, 2, 9, 0],
+          [0, 0, 6, 0, 1, 0, 2, 0, 0, 0, 9, 0],
+          [0, 0, 1, 1, 1, 0, 2, 0, 0, 0, 9, 0]
+        ]
+      },
       {
         "input": [
           [7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],

data/evaluation/d8e07eb2.json

@@ -35,9 +35,9 @@
         ],
         "output": [
           [3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3],
-        [3, 3, 0, 0, 0, 3, 3, 1, 1, 1, 3, 3, 7, 3, 3, 3, 3, 3, 6, 6, 3, 3],
+          [3, 3, 0, 0, 0, 3, 3, 2, 2, 3, 3, 3, 7, 3, 3, 3, 3, 3, 6, 6, 3, 3],
-        [3, 3, 0, 0, 0, 3, 3, 3, 1, 3, 3, 3, 7, 7, 7, 3, 3, 3, 3, 6, 3, 3],
+          [3, 3, 0, 0, 0, 3, 3, 2, 2, 2, 3, 3, 7, 7, 7, 3, 3, 3, 3, 6, 3, 3],
-        [3, 3, 0, 0, 0, 3, 3, 1, 1, 1, 3, 3, 7, 3, 3, 3, 3, 3, 6, 6, 3, 3],
+          [3, 3, 0, 0, 0, 3, 3, 3, 2, 3, 3, 3, 7, 3, 3, 3, 3, 3, 6, 6, 3, 3],
           [3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3],
           [6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6],
           [8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8],
@@ -465,4 +465,4 @@
         ]
       }
     ]
-}
+  }

readme.md

@@ -4,14 +4,14 @@ This repository contains the ARC-AGI-2 task data (ARC-AGI-1 can be found [here](
 
 *"ARC can be seen as a general artificial intelligence benchmark, as a program synthesis benchmark, or as a psychometric intelligence test. It is targeted at both humans and artificially intelligent systems that aim at emulating a human-like form of general fluid intelligence."*
 
-A foundational description of the dataset, its goals, and its underlying logic, can be found in: [On the Measure of Intelligence](https://arxiv.org/abs/1911.01547), the [ARC-AGI-2 Presentation](https://docs.google.com/presentation/d/1hQrGh5YI6MK3PalQYSQs4CQERrYBQZue8PBLjjHIMgI/edit?usp=sharing) and [ARC-AGI-2 Technical Report](http://arcprize.org/blog/arc-agi-2-technical-report)
+A foundational description of the dataset, its goals, and its underlying logic, can be found in: [On the Measure of Intelligence](https://arxiv.org/abs/1911.01547) and the [ARC-AGI-2 Presentation](https://docs.google.com/presentation/d/1hQrGh5YI6MK3PalQYSQs4CQERrYBQZue8PBLjjHIMgI/edit?usp=sharing)
 
 ## Dataset composition
 
-ARC-AGI-2 contains 1,000 public training tasks and 120 public evaluation tasks.
+ARC-AGI-2 contains 1,000 training tasks and 120 public evaluation tasks.
 
 The training tasks are intended to demonstrate the task format and the Core Knowledge priors used by ARC-AGI. They can be used for training AI models.
-The public evaluation tasks are intended for testing AI models that have never seen these tasks before. Average human performance on these tasks in our test sample was 66%.
+The public evaluation tasks are intended for testing AI models that have never seen these tasks before. Average human performance on these tasks in our test sample was 60%.
 
 ARC-AGI-2 also features two private test sets not included in the repo:
 
@@ -48,7 +48,7 @@ When looking at a task, a test-taker has access to inputs & outputs of the demon
 
 ## Usage of the testing interface
 
-You can view tasks on [ARCPrize.org/play](https://arcprize.org/play) or clone the [ARC-AGI-1 testing interface](https://github.com/fchollet/ARC-AGI/tree/master/apps). Open it in a web browser (Chrome recommended). It will prompt you to select a task JSON file.
+You can view tasks on [ARCPrize.org/play](https://arcprize.org/play) or clone the [ARC-AGI testing interface](https://github.com/fchollet/ARC-AGI/tree/master/apps) located at `apps/testing_interface.html`. Open it in a web browser (Chrome recommended). It will prompt you to select a task JSON file.
 
 After loading a task, you will enter the test space, which looks like this: