fix: switched up training and test splits in train.py

2 files changed, 0 insertions, 43 deletions

diff --git a/swr2_asr/utils/data.py b/swr2_asr/utils/data.py
index 10f0ea8..d551c98 100644
--- a/swr2_asr/utils/data.py
+++ b/swr2_asr/utils/data.py
@@ -134,11 +134,6 @@ class MLSDataset(Dataset):
 
     def initialize_limited(self) -> None:
         """Initializes the limited supervision dataset"""
-        # get file handles
-        # get file paths
-        # get transcripts
-        # create train or validation split
-
         handles = set()
 
         train_root_path = os.path.join(self.dataset_path, self.language, "train")
@@ -348,29 +343,3 @@ class MLSDataset(Dataset):
             dataset_lookup_entry["chapterid"],
             idx,
         )  # type: ignore
-
-
-if __name__ == "__main__":
-    DATASET_PATH = "/Volumes/pherkel/SWR2-ASR"
-    LANGUAGE = "mls_german_opus"
-    split = Split.DEV
-    DOWNLOAD = False
-
-    dataset = MLSDataset(DATASET_PATH, LANGUAGE, split, download=DOWNLOAD)
-
-    dataloader = DataLoader(
-        dataset,
-        batch_size=1,
-        shuffle=True,
-        collate_fn=DataProcessing(
-            "train", CharTokenizer.from_file("data/tokenizers/char_tokenizer_german.json")
-        ),
-    )
-
-    for batch in dataloader:
-        print(batch)
-        break
-
-    print(len(dataset))
-
-    print(dataset[0])
diff --git a/swr2_asr/utils/tokenizer.py b/swr2_asr/utils/tokenizer.py
index 22569eb..1cc7b84 100644
--- a/swr2_asr/utils/tokenizer.py
+++ b/swr2_asr/utils/tokenizer.py
@@ -120,15 +120,3 @@ class CharTokenizer:
                     load_tokenizer.char_map[char] = int(index)
                     load_tokenizer.index_map[int(index)] = char
         return load_tokenizer
-
-
-if __name__ == "__main__":
-    tokenizer = CharTokenizer.train("/Volumes/pherkel 1/SWR2-ASR", "mls_german_opus")
-    print(tokenizer.char_map)
-    print(tokenizer.index_map)
-    print(tokenizer.get_vocab_size())
-    print(tokenizer.get_blank_token())
-    print(tokenizer.get_unk_token())
-    print(tokenizer.get_space_token())
-    print(tokenizer.encode("hallo welt"))
-    print(tokenizer.decode([1, 2, 3, 4, 5, 6, 7, 8, 9, 10]))