Update dataset: 80K full corpus + 36K train/val/test splits (11 classes, clean)

full_corpus.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b9fb99099398843af4a807216b322cb15374ba6f3265a534e3ea08f0483a80d1
+size 31047554

test.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ae26e269c77a2cc97a8c5018ddbde4a87fc9a61cf2140b24b600fbd6608e947d
-size 2562293
+oid sha256:f9a820d98ca377684913c7566337ba986c97391c73c0ced223093d6838da641c
+size 1769027

train.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2949ff8ae4d43a5134f36f8be439cb6460936f4c8282f7433a86197e38d22a66
-size 21591456
+oid sha256:f9d0624183cbea998f65b615659b82342ec38893ad643e77de3c080bc79bb77d
+size 14018170

validation.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5f905ffb0432e37e507a8b99575534fa07def6bdb3af8565fa82cd388a4e389c
-size 2574833
+oid sha256:c21fc7700e85613b4e4ea3251876c016fb24a5db0ab57f8d0b6e97142fb14b58
+size 1743814