updates

examples/hunyuanvideo/hyvideo/dataset/text_dataset.py

@@ -44,6 +44,15 @@ def parse_data_file(self, data_file_path):
                 self.dataset = json.load(f)
         else:
             raise ValueError("Only support json and csv file now!")
+        assert len(self.dataset) > 0, "No data found in the data file."
+        sample = self.dataset[0]
+        assert (
+            self.caption_column in sample
+        ), f"Expected caption column `{self.caption_column}` in dataset, but got {sample.keys()}"
+
+        assert (
+            self.file_column in sample
+        ), f"Expected file path column {self.file_column} in dataset, but got {sample.keys()}"
 
     def __len__(self):
         return self.num_captions
@@ -63,13 +72,9 @@ def read_captions(self, dataset):
     def __getitem__(self, idx_text):
         idx = self.caption_sample_indices[idx_text]
         row = self.dataset[idx]
-        assert (
-            self.caption_column in row
-        ), f"Expected caption column {self.caption_column} in dataset, but got {row.keys()}"
         captions = row[self.caption_column]
         if isinstance(captions, str):
             captions = [captions]
-        assert self.file_column in row, f"Expected file path column {self.file_column} in dataset, but got {row.keys()}"
         file_path = row[self.file_column]
         # get the caption id
         first_text_index = self.caption_sample_indices.index(idx)

examples/hunyuanvideo/hyvideo/text_encoder/__init__.py

@@ -1,6 +1,7 @@
 from dataclasses import dataclass
 from typing import Optional, Tuple
 
+from hyvideo.constants import PRECISION_TO_TYPE, TEXT_ENCODER_PATH, TOKENIZER_PATH
 from hyvideo.utils.helpers import set_model_param_dtype
 from transformers import AutoTokenizer, CLIPTokenizer
 from transformers.utils import ModelOutput
@@ -12,8 +13,6 @@
 from mindone.transformers.models.llama.modeling_llama import ALL_LAYERNORM_LAYERS
 from mindone.utils.amp import auto_mixed_precision
 
-from constants import PRECISION_TO_TYPE, TEXT_ENCODER_PATH, TOKENIZER_PATH
-
 
 def use_default(value, default):
     return value if value is not None else default

examples/hunyuanvideo/scripts/run_text_encoder.py

@@ -201,8 +201,8 @@ def parse_args():
     )
     parser.add_argument(
         "--file-column",
-        default="path",
-        help="The column of file path in `data_file_path`. Defaults to `path`.",
+        default="video",
+        help="The column of file path in `data_file_path`. Defaults to `video`.",
     )
     parser.add_argument(
         "--caption-column",
@@ -309,6 +309,7 @@ def main(args):
     print(f"rank_id {rank_id}, device_num {device_num}")
 
     # build dataloader for large amount of captions
+    print_banner("data init")
     if args.data_file_path is not None:
         assert isinstance(args.data_file_path, str), "Expect data_file_path to be a string!"
         assert Path(args.data_file_path).exists(), "data_file_path does not exist!"