Update README.md

Sample traces to train VLMs for enhanced spatial reasoning using R1-style CoT for more robust distance estimates grounded on measurements from a 3D scene reconstruction pipeline.

See more in https://huggingface.co/spaces/open-r1/README/discussions/10

README.md

@@ -19,13 +19,22 @@ dataset_info:
     dtype: string
   splits:
   - name: train
-    num_bytes: 2166206.0
+    num_bytes: 2166206
     num_examples: 20
   download_size: 2143240
-  dataset_size: 2166206.0
+  dataset_size: 2166206
 configs:
 - config_name: default
   data_files:
   - split: train
     path: data/train-*
----
+task_categories:
+- visual-question-answering
+tags:
+- reasoning
+- spatial
+- vqa
+- remyx
+size_categories:
+- n<1K
+---