Support Multimodal datasets

Author: nikita-smetanin

src/together/utils/files.py

@@ -377,7 +377,8 @@ def _check_message_content(
         return True, num_images
 
     raise InvalidFileFormatError(
-        message=f"Invalid content type on line {idx + 1} of the input file. Found {type(message_content)}",
+        f"Invalid content type on line {idx + 1} of the input file. Expected string or multimodal list of dicts, "
+        f"found {type(message_content)}",
         line_number=idx + 1,
         error_source="key_value",
     )

tests/unit/test_files_checks.py

@@ -1,10 +1,8 @@
-import json
-import pytest
 import csv
+import json
 from pathlib import Path
 
-from together.constants import MIN_SAMPLES
-from together.utils.files import check_file, FilePurpose
+from together.utils.files import FilePurpose, check_file
 
 
 def test_check_jsonl_valid_general(tmp_path: Path):