Fix OCR misspellings; drop 3 duplicate math Qs (2069->2066); update counts

README.md

@@ -19,19 +19,19 @@ dataset_info:
     dtype: large_string
   splits:
   - name: tarix
-    num_bytes: 298902
-    num_examples: 809
+    num_bytes: 178316
+    num_examples: 509
   - name: ona_tili
-    num_bytes: 370072
-    num_examples: 826
+    num_bytes: 175489
+    num_examples: 426
   - name: fizika
-    num_bytes: 169945
-    num_examples: 554
+    num_bytes: 108135
+    num_examples: 404
   - name: matematika
-    num_bytes: 212688
-    num_examples: 880
-  download_size: 456205
-  dataset_size: 1051607
+    num_bytes: 156202
+    num_examples: 727
+  download_size: 302047
+  dataset_size: 618142
 configs:
 - config_name: default
   data_files:
@@ -44,3 +44,22 @@ configs:
   - split: matematika
     path: data/matematika-*
 ---
+
+# DTM2019 (public release)
+
+Uzbek-language multiple-choice questions extracted from DTM 2019 university-entrance
+exam preparation materials. This **public release contains 2,066 questions**.
+
+It is the full 3,066-question corpus with the **1,000-question evaluation benchmark
+subset held out and not published** (used for model evaluation at eval.idrock.uz).
+
+| Subject | Questions |
+|---|---|
+| Mathematics (matematika) | 727 |
+| Physics (fizika) | 404 |
+| Uzbek language (ona_tili) | 426 |
+| History (tarix) | 509 |
+| **Total** | **2,066** |
+
+Each record: `test_number`, `topic`, `question`, `option_A`–`option_D`, `answer`.
+Mathematical content is in LaTeX; underlined fragments in Ona tili items use `<u>...</u>`.

data/fizika-00000-of-00001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5a8ddedef1f00cbde49948794e7a56acd75efdb0094ecbe080d113c7c50aee91
-size 60936
+oid sha256:50ab35e7fc7d515e184a3684c1050031dfe3a690f2055e91e2cfce118b299130
+size 45090

data/matematika-00000-of-00001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b75aeba7e8098a76ae4a8b104868f3f04dc63f674f6bf9557144c2702dd011ae
-size 80160
+oid sha256:ae56ecf2857fe0328237f5060b9a6c5c438c404714be9ba192a008af79976476
+size 66807

data/ona_tili-00000-of-00001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:15bc7bd2e223ebd2d7f926c9266fe385a7e721d04c030bbf080232a6a79a6384
-size 182132
+oid sha256:325374a7c4a1f852986d93e36138523ba2e35c32664c5c5ada49c84bd79f71c7
+size 101181

data/tarix-00000-of-00001.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:47d96dc3e0882c493d078d2408f534120e66c09a3cb8ceb91906e5fd350534fb
-size 132977
+oid sha256:b35598b8ff378ad4f2532ce7502f0ff54848851f3e0459b904ed509e0585688c
+size 88969