Schema migrate mm_my: gt_transcript→human_transcript + canonical 18-col (batch 41-43)

data/mm_my/train-00037.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1d37964b8dcb4cbee6d5eb00de21126b56db42d6be0dbc472f55d482c71116e3
-size 1716048
+oid sha256:a79c7a2cf645b579b32c1ba604356330435e09bfeaa3f91cfb25a6cbccb5c0f0
+size 1707594

data/mm_my/train-00038.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0663bd047a20806ff72366814b29f0a048d4cefa7e515ad61f0eccbcd66176fe
-size 1703637
+oid sha256:920a62082b393ed54c52a10ba61fb1780722cff8aa66d5bb24bbbd9de4ae52ef
+size 1692270

data/mm_my/train-00039.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:51f63b66e40b35afc56dafb0b3c3ecf8b68f8219aedae8ae59d2c5999ae86398
-size 701858
+oid sha256:b866bcffb7bcd2190f344b3a8632cd4969294293cda81a815b57721bb1e3e602
+size 698411