remove overused warnings & fix wording & prevent un-matchable models and datasets submissions

app_text_classification.py

@@ -2,7 +2,7 @@ import uuid
 
 import gradio as gr
 
-from io_utils import get_logs_file, read_scanners, write_scanners
+from io_utils import read_scanners, write_scanners
 from text_classification_ui_helpers import (
     get_related_datasets_from_leaderboard,
     align_columns_and_show_prediction,
@@ -11,7 +11,7 @@ from text_classification_ui_helpers import (
     try_submit,
     write_column_mapping_to_config,
 )
-from wordings import CONFIRM_MAPPING_DETAILS_MD, INTRODUCTION_MD, USE_INFERENCE_API_TIP
+from wordings import CONFIRM_MAPPING_DETAILS_MD, INTRODUCTION_MD, USE_INFERENCE_API_TIP, CHECK_LOG_SECTION_RAW
 
 MAX_LABELS = 40
 MAX_FEATURES = 20
@@ -114,8 +114,8 @@ def get_demo():
 
     with gr.Row():
         logs = gr.Textbox(
-            value=get_logs_file,
-            label="Giskard Bot Evaluation Log:",
+            value=CHECK_LOG_SECTION_RAW,
+            label="Giskard Bot Evaluation Guide:",
             visible=False,
             every=0.5,
         )
@@ -135,7 +135,7 @@ def get_demo():
     )
     
     gr.on(
-        triggers=[dataset_id_input.input],
+        triggers=[dataset_id_input.change],
         fn=check_dataset,
         inputs=[dataset_id_input],
         outputs=[dataset_config_input, dataset_split_input, loading_status]

text_classification_ui_helpers.py

@@ -23,6 +23,7 @@ from wordings import (
     CONFIRM_MAPPING_DETAILS_FAIL_RAW,
     MAPPING_STYLED_ERROR_WARNING,
     NOT_TEXT_CLASSIFICATION_MODEL_RAW,
+    UNMATCHED_MODEL_DATASET,
     get_styled_input,
 )
 
@@ -41,7 +42,7 @@ def get_related_datasets_from_leaderboard(model_id):
     if len(datasets_unique) == 0:
         return gr.update(choices=[], value="")
     
-    return gr.update(choices=datasets_unique, value=datasets_unique[0])
+    return gr.update(choices=datasets_unique, value="")
 
 
 logger = logging.getLogger(__file__)
@@ -185,7 +186,7 @@ def precheck_model_ds_enable_example_btn(
         return (gr.update(interactive=True), gr.update(value=df, visible=True), "")
     except Exception as e:
         # Config or split wrong
-        gr.Warning(f"Failed to load dataset {dataset_id} with config {dataset_config}: {e}")
+        logger.warn(f"Check your dataset {dataset_id} and config {dataset_config} on split {dataset_split}: {e}")
         return (gr.update(interactive=False), gr.update(value=pd.DataFrame(), visible=False), "")
 
 
@@ -255,6 +256,17 @@ def align_columns_and_show_prediction(
             "",
             *dropdown_placement,
         )
+    
+    if len(ds_labels) != len(model_labels):
+        gr.Warning(UNMATCHED_MODEL_DATASET)
+        return (
+            gr.update(visible=False),
+            gr.update(visible=False),
+            gr.update(visible=False, open=False),
+            gr.update(interactive=False),
+            "",
+            *dropdown_placement,
+        )
 
     column_mappings = list_labels_and_features_from_dataset(
         ds_labels,
@@ -301,10 +313,10 @@ def check_column_mapping_keys_validity(all_mappings):
 def construct_label_and_feature_mapping(all_mappings, ds_labels, ds_features):
     label_mapping = {}
     if len(all_mappings["labels"].keys()) != len(ds_labels):
-        gr.Warning("Label mapping corrupted: " + CONFIRM_MAPPING_DETAILS_FAIL_RAW)
+        logger.warn("Label mapping corrupted: " + CONFIRM_MAPPING_DETAILS_FAIL_RAW)
       
     if len(all_mappings["features"].keys()) != len(ds_features):
-        gr.Warning("Feature mapping corrupted: " + CONFIRM_MAPPING_DETAILS_FAIL_RAW)
+        logger.warn("Feature mapping corrupted: " + CONFIRM_MAPPING_DETAILS_FAIL_RAW)
 
     for i, label in zip(range(len(ds_labels)),  ds_labels):
         # align the saved labels with dataset labels order

wordings.py

@@ -17,14 +17,22 @@ CONFIRM_MAPPING_DETAILS_FAIL_MD = """
                             Sorry, we cannot align the input/output of your dataset with the model. <b>Pleaser double check your model and dataset.</b>
                             """
 
+UNMATCHED_MODEL_DATASET = """
+                            Model prediction labels do not align with the labels present in the dataset. Please double check your model and dataset.
+                            """
+
 CONFIRM_MAPPING_DETAILS_FAIL_RAW = """
-                            Sorry, we cannot align the input/output of your dataset with the model. Pleaser double check your model and dataset.
+                            Sorry, we cannot auto-align the input/output of your dataset with the model.
                             """
 
 CHECK_CONFIG_OR_SPLIT_RAW = """
                             Please check your dataset config or split.
                             """
 
+CHECK_LOG_SECTION_RAW = """
+                          Your evaluation job has been successfully submitted. Further details are available in the Logs tab, providing information on your queue status and the current job log.
+                        """
+
 PREDICTION_SAMPLE_MD = """
                         <h1 style="text-align: center;">
                         Model Prediction Sample