Adding intermediate format, where JSONL is present, but we also use our list-based sentence structure for each text.

data/test.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d2df9402544def04187f8a0eb32719f6c28969e66b57d0f736e6206b07f87c04
-size 9136558
+oid sha256:ea8749b628b7976384d31f3b5c091666ecb55e79d40109ba820e6954d5dacf25
+size 9336683

data/train.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2b7830aa8b2d73d9bfbda9c9ddecf5543e778ff6acc96895f4c39a8d21afd50e
-size 91417326
+oid sha256:da3a49947115282602a5dab889953472c8c29370e7e1091eb35d9d64a67ab5bb
+size 93382931

data/validation.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9f96938795e8c7f223e22c6dede003d209ac0847e51fac55de83e111ba256d1b
-size 9153979
+oid sha256:1e04d49112233742b960567c1184165b635b2057e65840eadc09fde8713d787e
+size 9355777