clean: more correect_language exceptions

Author: ZJaume

30.clean

@@ -28,9 +28,9 @@ OUTPUT=$WORKSPACE/$output_dir/$L/${L}_$SLURM_ARRAY_TASK_ID.jsonl.zst
 mkdir -p $WORKSPACE/$output_dir/$L
 
 case "$L" in
-    zh | ja | ko | th)
+    zh | ja | ko)
         FILTER_PARAMS="-a -z";;
-    ms | id | uz | sw | so)
+    af | tl | ms | id | uz | sw | so)
         FILTER_PARAMS="-e -w -m";;
     fi)
         FILTER_PARAMS="-a -E";;