Release 25 I2T and 7 I2I task splits

README.md

@@ -22,8 +22,12 @@ configs:
     path: data/recognition__anomaly_detection__i2t/val/*.parquet
   - split: recognition__appearance_understanding__i2t
     path: data/recognition__appearance_understanding__i2t/val/*.parquet
+  - split: recognition__attribute_editing__i2i
+    path: data/recognition__attribute_editing__i2i/train/*.parquet
   - split: recognition__category_recognition__i2t
     path: data/recognition__category_recognition__i2t/val/*.parquet
+  - split: recognition__object_editing__i2i
+    path: data/recognition__object_editing__i2i/train/*.parquet
   - split: recognition__ocr_and_text_recognition__i2t
     path: data/recognition__ocr_and_text_recognition__i2t/val/*.parquet
   - split: recognition__part_recognition__i2t
@@ -44,6 +48,8 @@ configs:
     path: data/reconstruction__3d_shape_reasoning__i2t/val/*.parquet
   - split: reconstruction__depth_understanding__i2t
     path: data/reconstruction__depth_understanding__i2t/val/*.parquet
+  - split: reconstruction__inpainting__i2i
+    path: data/reconstruction__inpainting__i2i/train/*.parquet
   - split: reconstruction__lighting_understanding__i2t
     path: data/reconstruction__lighting_understanding__i2t/val/*.parquet
   - split: reconstruction__metric_3d_relation__i2t
@@ -62,8 +68,16 @@ configs:
     path: data/reorganization__connectivity__i2t/val/*.parquet
   - split: reorganization__counting__i2t
     path: data/reorganization__counting__i2t/val/*.parquet
+  - split: reorganization__jigsaw__i2i
+    path: data/reorganization__jigsaw__i2i/train/*.parquet
+  - split: reorganization__localization__i2i
+    path: data/reorganization__localization__i2i/train/*.parquet
   - split: reorganization__localization__i2t
     path: data/reorganization__localization__i2t/val/*.parquet
+  - split: reorganization__object_pointing__i2i
+    path: data/reorganization__object_pointing__i2i/train/*.parquet
+  - split: reorganization__semantic_segmentation__i2i
+    path: data/reorganization__semantic_segmentation__i2i/train/*.parquet
   - split: reorganization__visual_correspondence__i2t
     path: data/reorganization__visual_correspondence__i2t/val/*.parquet
 ---
@@ -74,7 +88,7 @@ OmniTaskonomy groups visual tasks into Recognition, Reconstruction, and Reorgani
 Each task and modality has its own split, named `family__task__i2i` or `family__task__i2t`.
 
 This release contains **9,444 I2T evaluation samples across 25 tasks**
-and **0 I2I training samples across 0 tasks**.
+and **350,000 I2I training samples across 7 tasks**.
 I2T rows have `usage="val"`; I2I rows have `usage="train"`.
 I2I validation data is not included.
 Counts and file checksums are in [release_manifest.json](release_manifest.json).
@@ -105,11 +119,15 @@ Original I2T image bytes, image order, questions, and answers are preserved;
 `evaluation_prompt` retains the benchmark's evaluation wording, including its native CV-Bench prompt.
 Source annotation anomalies are retained.
 
-I2I targets use the training reader's rendering, including Taskonomy target normalization.
+I2I targets use the training reader's rendering.
+For damaged source files accepted by the training reader, the same decoded pixels are saved
+as valid PNGs; affected records are identified in `source_metadata`.
 Localization repeats the original training iterator to 50,000 records.
 Object pointing stores one deterministic choice from the training prompt templates;
 the templates remain available in source metadata.
 
+Taskonomy-derived I2I splits are not included while redistribution permission is pending.
+
 ## License
 
 This dataset combines sources with different licenses. See [LICENSE.md](LICENSE.md)