Append 14,111 full training rows; preserve published data

CHANGELOG.md

@@ -4,6 +4,12 @@ Changes that affect published data, newest first: one entry per publish, `## <da
 concise bullets on what changed in the data. `tasksource-jev-typed-decisions` releases are tagged on the Hub
 (`load_dataset("tasksource/tasksource-jev-typed-decisions", revision="v2026.10.13")`).
 
+## tasksource-jev-typed-decisions v2026.10.17.1
+
+- `full`: add EMS-MCQA’s native single-answer emergency-medical practice questions, preserving all options, including five- and six-choice questions.
+- Exclude multi-answer records, malformed or duplicate options and positional references that cannot survive option shuffling. Retain native source splits; exclude collection-wide aliases from loading into the task.
+- Keep source website, certification/category and rationale in metadata, separate from model inputs. `default` receives the task at its next full rebuild.
+
 ## tasksource-jev-typed-decisions v2026.10.17
 
 - `vision`: add ShapeWorld object-box grounding, capped at 20k training rows from the 100k-QA procedural mirror. Native scene models verify every unique target and spatial relation.

README.md

@@ -161,16 +161,16 @@ dataset_info:
     dtype: string
   splits:
   - name: train
-    num_bytes: 3836028641
-    num_examples: 2500000
+    num_bytes: 3844158050
+    num_examples: 2514111
   - name: validation
     num_bytes: 23419061
     num_examples: 15000
   - name: test
     num_bytes: 21143086
     num_examples: 15000
-  download_size: 1935206000
-  dataset_size: 3880590788
+  download_size: 1939084021
+  dataset_size: 3888720197
 - config_name: vision
   features:
   - name: images

full/additions/517726f3090c/manifest.json

@@ -0,0 +1,42 @@
+{
+  "config": "full",
+  "baseline_revision": "b965b6659ee5ca90fbf99a38238c87bb95c29b5d",
+  "seed": 0,
+  "max_rows": 20000,
+  "added_rows": 14111,
+  "sources": {
+    "ems-mcqa": 14111
+  },
+  "dropped_overlap": {
+    "built": 14112,
+    "train": 1,
+    "evaluation": 0
+  },
+  "revisions": {
+    "ems-mcqa": {
+      "Xueren/EMS-MCQA": "dd0a05836786cb8c2de489916b13505531323d8e"
+    }
+  },
+  "evaluation": "unchanged",
+  "default": "unchanged until the next full rebuild",
+  "preserved_parquets": {
+    "full/test-00000-of-00001.parquet": "f2ab059f5a85de7666b1a2d3b443db292b91f854",
+    "full/train-00000-of-00015.parquet": "f760ee18afd4ce70b015751ef79fc6fd818eae22",
+    "full/train-00001-of-00015.parquet": "bdf7e70d4ab8cdd55500fdb881d37b0121132510",
+    "full/train-00002-of-00015.parquet": "bc1a1bc64592bcf967676340974dac5430258b17",
+    "full/train-00003-of-00015.parquet": "35387df68c882a632c5db1b0640b6c35a20a3db1",
+    "full/train-00004-of-00015.parquet": "41b77b9438781e13b1d5170b7b478b7171023e2c",
+    "full/train-00005-of-00015.parquet": "2aca50a156ee0e00e694a89f512e0adaa06115b3",
+    "full/train-00006-of-00015.parquet": "59e41fe02dffc640932fb77c1cdf9dd0e02318c7",
+    "full/train-00007-of-00015.parquet": "351a3a06f2b67a15bd5e09204a0d56a7db6b2231",
+    "full/train-00008-of-00015.parquet": "ac056cd110fb57910182062b43e9ad22d225dce0",
+    "full/train-00009-of-00015.parquet": "9cafbac20e9cd9c6c2370dcf4a1d7e85b7d57a50",
+    "full/train-00010-of-00015.parquet": "6cf74222bb104a4b6ae168a0f92dfc950184af9b",
+    "full/train-00011-of-00015.parquet": "b15741e8207198c5b6be6b9bd589f039ddc5c6c9",
+    "full/train-00012-of-00015.parquet": "fac82ce7720e4476dfd75761e66b72cd99544c12",
+    "full/train-00013-of-00015.parquet": "72aaac7b41d430853a0d55a9d0f9356fbd0d4b56",
+    "full/train-00014-of-00015.parquet": "b1453c5e9bd1805fb7bfdda3fd3b55e4debfccd8",
+    "full/validation-00000-of-00001.parquet": "96211364bc43a2349890d3aa9714d4b23bfaf2ba"
+  },
+  "script_sha256": "d99649747f8c3b5cb39ebc7faa179ce4cc2148193c88595b1e4c6002b782b58b"
+}

full/train-append-517726f3090c-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:16c66560d23861cd783497e30acfc759a2e33575b1853b0e219806573b45796d
+size 3878021

release-audit.json

@@ -8,7 +8,7 @@
   },
   "splits": {
     "train": {
-      "rows": 2500000,
+      "rows": 2514111,
       "sources": {
         "AES2-essay-scoring": 2521,
         "AdjectiveScaleProbe-nli": 3995,
@@ -760,11 +760,12 @@
         "wiqa": 12125,
         "wnut_17/wnut_17": 7389,
         "yahoo_answers_topics": 5119,
-        "yelp_review_full/yelp_review_full": 4198
+        "yelp_review_full/yelp_review_full": 4198,
+        "ems-mcqa": 14111
       },
       "formats": {
         "Classification": 1198861,
-        "MultipleChoice": 765230,
+        "MultipleChoice": 779341,
         "graded": 250000,
         "procedural": 250000,
         "TokenClassification": 35909
@@ -1199,23 +1200,24 @@
         "wiqa": 12125,
         "wnut_17": 7389,
         "yahoo_answers_topics": 5119,
-        "yelp_review_full": 4198
+        "yelp_review_full": 4198,
+        "ems-mcqa": 14111
       },
       "kinds": {
-        "choice": 1963397,
-        "noul": 294083,
+        "choice": 1976836,
+        "noul": 294755,
         "score": 242520
       },
       "variants": {
-        "criteria_permutation": 79765,
-        "direct": 2159106,
+        "criteria_permutation": 80400,
+        "direct": 2171910,
         "instruction_paraphrase": 32805,
-        "label_verification": 79979,
+        "label_verification": 80651,
         "packed_derived": 124140,
         "paired_text_format": 24205
       },
       "license_use": {
-        "commercial": 1393330,
+        "commercial": 1407441,
         "non-commercial": 270750,
         "unspecified": 835920
       }
@@ -3558,5 +3560,11 @@
       "rows": 15000,
       "sources": 715
     }
-  }
+  },
+  "incremental_additions": [
+    {
+      "manifest": "full/additions/517726f3090c/manifest.json",
+      "train_rows": 14111
+    }
+  ]
 }

sources.yaml

@@ -68,6 +68,7 @@ datasets:
 - WhereIsAI/github-issue-similarity
 - WillHeld/top_v2
 - X-LANCE/WebSRC_v1.0
+- Xueren/EMS-MCQA
 - Yelp/yelp_review_full
 - YufeiHFUT/DocRED_origin
 - Zhengping/UNLI
@@ -10449,3 +10450,24 @@ sources:
     card_licenses:
       Yelp/yelp_review_full:
       - other
+  ems-mcqa:
+    rows:
+      train: 14111
+    dataset: Xueren/EMS-MCQA
+    revision: dd0a05836786cb8c2de489916b13505531323d8e
+    revisions:
+      Xueren/EMS-MCQA: dd0a05836786cb8c2de489916b13505531323d8e
+    license: mit, mit (original source)
+    license_use: commercial
+    card_licenses:
+      Xueren/EMS-MCQA:
+      - mit
+    source_license_evidence:
+      license: mit
+      license_use: commercial
+      checked: '2026-10-17'
+      scope: Publisher-declared MIT for the released public practice-question subset; source websites remain identified per
+        row. Expert review covered 100 questions, not every training label.
+      url: https://huggingface.co/datasets/Xueren/EMS-MCQA
+      additional_urls:
+      - https://arxiv.org/html/2511.10900v1