Update load_data.py

load_data.py

@@ -11,6 +11,7 @@ from argilla.listeners import listener
 
 HF_TOKEN = os.environ.get("HF_TOKEN")
 HUB_DATASET_NAME = "mserras/alpaca-es-hackaton"
+HUB_DATASET_NAME_VAL = "mserras/alpaca-es-hackaton-validated"
 
 @listener(
     dataset="somos-alpaca-es", 
@@ -23,7 +24,7 @@ def save_validated_to_hub(records, ctx):
         if HF_TOKEN:
             print("Pushing the dataset")
             print(ds)
-            ds.push_to_hub(HUB_DATASET_NAME, token=HF_TOKEN)
+            ds.push_to_hub(HUB_DATASET_NAME_VAL, token=HF_TOKEN)
         else:
             print("SET HF_TOKEN and HUB_DATASET_NAME TO SYNC YOUR DATASET!!!")
     else:
@@ -70,7 +71,7 @@ class LoadDatasets:
         rg.log(
             records,
             name="somos-alpaca-es",
-            tags={"description": "SomosNLP Hackathon dataset"},
+            tags={"description": "SomosNLP Hackathon dataset - instruction filtering version"},
             batch_size=200
         )