Upload README.md with huggingface_hub

README.md

@@ -33,4 +33,32 @@ configs:
     path: data/train-*
   - split: validation
     path: data/validation-*
+language:
+- en
+license: cc-by-4.0
+source_datasets:
+- maveryn/trace
+task_categories:
+- multiple-choice
+- visual-question-answering
+tags:
+- decision-model
+- procedural
 ---
+
+# trace-mcq
+
+TRACE static MC reuses original rendered images and native train/validation rows.
+
+Source: https://huggingface.co/datasets/maveryn/trace. Distractors are answers
+from other native training instances of the same task and answer type, rejected
+by TRACE's exact-JSON answer scorer for the current example. No annotation targets,
+JSON demonstrations or execution traces enter model inputs. Native labels and image
+bytes are preserved; no new TRACE generation or model-derived labels are needed.
+Dataset: CC BY 4.0; verifier code: Apache-2.0; original asset notices remain applicable.
+
+Original data: [maveryn/trace](https://huggingface.co/datasets/maveryn/trace), [original distribution](https://huggingface.co/datasets/maveryn/trace). Repackaged as parquet for [tasksource](https://github.com/sileod/tasksource) by [scripts/repackage_dataset/procedural/trace.py](https://github.com/sileod/tasksource/blob/main/scripts/repackage_dataset/procedural/trace.py#L49).
+
+[Conversion provenance](provenance.json): source revisions, hashes, configuration and exclusions.
+
+[Upstream asset credits and licenses](asset-notices.json) accompany the original rendered images.