diff --git a/model/model_training/custom_datasets/qa_datasets.py b/model/model_training/custom_datasets/qa_datasets.py
index ed778ea09b..20f5ea851f 100644
--- a/model/model_training/custom_datasets/qa_datasets.py
+++ b/model/model_training/custom_datasets/qa_datasets.py
@@ -275,32 +275,17 @@ def __getitem__(self, index) -> list[str] | tuple[str]:
             return (self.pairs[index][0] + " " + self.pairs[index][1],)
 
 
-class SODADialogue(Dataset):
-    url = "https://drive.google.com/uc?id=1TOGQfr419n8wpzJpYLLw4nB3tSKD8zXV"
-
+class SODADialogue:
     def __init__(self, cache_dir, verbose=True):
-        path = os.path.join(cache_dir, "soda_dialog.jsonl")
-
-        if not os.path.exists(path):
-            import gzip
-            import shutil
-
-            import gdown
-
-            gdown.download(self.url, output=os.path.join(cache_dir, "soda_dialog.jsonl.gz"))
-
-            with gzip.open(os.path.join(cache_dir, "soda_dialog.jsonl.gz"), "rb") as f_in:
-                with open(path, "wb") as f_out:
-                    shutil.copyfileobj(f_in, f_out)
+        dataset = load_dataset("emozilla/soda_synthetic_dialogue", cache_dir=cache_dir)
 
         self.pairs = []
         faulty = 0
-        with open(path) as fin:
-            for line in fin:
-                conversation = json.loads(line)
+        for split in dataset:
+            for row in dataset[split]:
                 question_answer_pairs = ()
 
-                question_answers = conversation["text"].split("User: ")
+                question_answers = row["conversation"].split("User: ")
                 for question_answer in question_answers[1:]:  # first element is empty
                     try:
                         question, answer = question_answer.split("\nAssistant: ")