Fix pretraining with iterable/streaming Dataset (#556)

* return without packing prep/len

* fix remove columns

* fix encode arguments

* add error when max steps not set

* fix test

---------

Co-authored-by: Jan Philipp Harries <jphme@users.noreply.github.com>

tests/test_data.py

@@ -35,7 +35,7 @@ class TestEncodePretraining(unittest.TestCase):
                 "hello, hello",
             ]
         }
-        result = encode_pretraining(self.tokenizer, self.max_tokens, examples)
+        result = encode_pretraining(self.tokenizer, self.max_tokens, examples["text"])
 
         self.assertEqual(len(result["input_ids"]), 3)