Add tasksource-jev-typed-decisions dataset card

README.md

@@ -16,51 +16,6 @@ tags:
 - multiple-choice
 size_categories:
 - 1M<n<10M
-configs:
-- config_name: default
-  data_files:
-  - split: train
-    path: data/train-*
-  - split: validation
-    path: data/validation-*
-  - split: test
-    path: data/test-*
-dataset_info:
-  features:
-  - name: state
-    dtype: string
-  - name: id
-    dtype: string
-  - name: kind
-    dtype: string
-  - name: options
-    list: string
-  - name: target
-    list: float64
-  - name: question
-    dtype: string
-  - name: source
-    dtype: string
-  - name: variant
-    dtype: string
-  - name: split
-    dtype: string
-  - name: group_id
-    dtype: string
-  - name: question_id
-    dtype: string
-  splits:
-  - name: train
-    num_bytes: 1219014177
-    num_examples: 1000000
-  - name: validation
-    num_bytes: 18091829
-    num_examples: 15000
-  - name: test
-    num_bytes: 19146132
-    num_examples: 15000
-  download_size: 707871258
-  dataset_size: 1256252138
 ---
 
 # tasksource-jev-typed-decisions
@@ -142,6 +97,7 @@ following each source's own train/dev/test splits where it has them.
 - **Mixing.** Formats get fixed shares of the train rows (47% classification, 30% multiple choice, 3% token labeling, 10% graded, 10% procedural). Within a format, dataset families get equal shares, scaled by [hand-set weights](https://github.com/sileod/tasksource/blob/main/src/tasksource/metadata/weights.py) (more for adversarial NLI, long documents and preference pairs; less for templated probes). Related questions are kept together.
   - The first 1,000 train rows are interleaved to show variety in the Dataset Viewer; the rest is shuffled.
   - Evaluation benchmarks (BIG-bench, MMLU, BLiMP, MATH test, ...) are left out so they stay clean for evaluation.
+- **Sources.** [sources.yaml](sources.yaml) lists every source with its rows, the Hub dataset and revision it was loaded from, and the original dataset behind each tasksource copy.
 - **Audit trail.** The [source mix](release-audit.json), [failed source list](failed-tasks.json), and [build manifest](build-manifest.json) ship with the data.
 - **Reproducible.** The [build runbook](https://github.com/sileod/tasksource/blob/main/docs/jev/README.md) rebuilds the release from [Tasksource](https://github.com/sileod/tasksource)'s [task catalog](https://github.com/sileod/tasksource/blob/main/tasks.md).