add labelling

src/synthetic_dataset_generator/apps/base.py

@@ -79,8 +79,10 @@ def validate_push_to_hub(org_name, repo_name):
 
 def combine_datasets(repo_id: str, dataset: Dataset) -> Dataset:
     try:
-        dataset = load_dataset(repo_id, split="train")
-        return concatenate_datasets([dataset, dataset])
+        new_dataset = load_dataset(
+            repo_id, split="train", download_mode="force_redownload"
+        )
+        return concatenate_datasets([dataset, new_dataset])
     except Exception:
         return dataset
 

src/synthetic_dataset_generator/apps/textcat.py

@@ -105,7 +105,7 @@ def generate_dataset(
         is_sample=is_sample,
     )
     labeller_generator = get_labeller_generator(
-        system_prompt=f"{system_prompt}. Optional labels: {', '.join(labels)}. Only select actually relevant labels.",
+        system_prompt=f"{system_prompt}. Optional labels: {', '.join(labels)}. Only apply relevant labels. Applying less labels is better than applying too many labels.",
         labels=labels,
         multi_label=multi_label,
     )