add suggested dataset

app_leaderboard.py

@@ -7,6 +7,7 @@ from fetch_utils import (check_dataset_and_get_config,
                          check_dataset_and_get_split)
 from text_classification_ui_helpers import LEADERBOARD
 
+import leaderboard
 
 def get_records_from_dataset_repo(dataset_id):
     dataset_config = check_dataset_and_get_config(dataset_id)
@@ -74,7 +75,8 @@ def get_display_df(df):
 
 
 def get_demo():
-    records = get_records_from_dataset_repo(LEADERBOARD)
+    leaderboard.records = get_records_from_dataset_repo(LEADERBOARD)
+    records = leaderboard.records
 
     model_ids = get_model_ids(records)
     dataset_ids = get_dataset_ids(records)
@@ -124,6 +126,7 @@ def get_demo():
         outputs=[leaderboard_df],
     )
     def filter_table(model_id, dataset_id, columns, task):
+        records = leaderboard.records
         # filter the table based on task
         df = records[(records["task"] == task)]
         # filter the table based on the model_id and dataset_id

app_text_classification.py

@@ -4,6 +4,7 @@ import gradio as gr
 
 from io_utils import get_logs_file, read_scanners, write_scanners
 from text_classification_ui_helpers import (
+    get_related_datasets_from_leaderboard,
     align_columns_and_show_prediction,
     check_dataset,
     deselect_run_inference,
@@ -34,10 +35,13 @@ def get_demo():
             placeholder=EXAMPLE_MODEL_ID + " (press enter to confirm)",
         )
 
-        dataset_id_input = gr.Textbox(
-            label="Hugging Face Dataset id",
-            placeholder=EXAMPLE_DATA_ID + " (press enter to confirm)",
-        )
+        with gr.Column():
+            no_dataset_checkbox = gr.Checkbox(label="Recommend a dataset", value=False, visible=True)
+
+            dataset_id_input = gr.Textbox(
+                label="Hugging Face Dataset id",
+                placeholder=EXAMPLE_DATA_ID + " (press enter to confirm)",
+            )
 
     with gr.Row():
         dataset_config_input = gr.Dropdown(label="Dataset Config", visible=False, allow_custom_value=True)
@@ -149,6 +153,13 @@ def get_demo():
         outputs=[inference_token, run_inference],
     )
 
+    gr.on(
+        triggers=[model_id_input.change, no_dataset_checkbox.change],
+        fn=get_related_datasets_from_leaderboard,
+        inputs=[model_id_input, no_dataset_checkbox],
+        outputs=[dataset_id_input],
+    )
+
     gr.on(
         triggers=[label.change for label in column_mappings],
         fn=write_column_mapping_to_config,

leaderboard.py

@@ -0,0 +1,3 @@
+import pandas as pd
+
+records = pd.DataFrame()

text_classification_ui_helpers.py

@@ -4,6 +4,7 @@ import logging
 import os
 import threading
 import uuid
+import leaderboard
 
 import datasets
 import gradio as gr
@@ -42,6 +43,17 @@ HF_GSK_HUB_HF_TOKEN = "GSK_HF_TOKEN"
 HF_GSK_HUB_UNLOCK_TOKEN = "GSK_HUB_UNLOCK_TOKEN"
 
 LEADERBOARD = "giskard-bot/evaluator-leaderboard"
+def get_related_datasets_from_leaderboard(model_id, rec_dataset=False):
+    if not rec_dataset:
+        return gr.Textbox(placeholder="Please enter your dataset id.", visible=True)
+    records = leaderboard.records
+    model_records = records[records["model_id"] == model_id]
+    print(model_records)
+    datasets_unique = model_records["dataset_id"].unique()
+    print(datasets_unique)
+    if len(datasets_unique) == 0:
+        return gr.Textbox(placeholder="Sorry, we don't have recommendations for this model yet.", visible=True)
+    return gr.Textbox(label="Recommended Datasets", value=datasets_unique[0], visible=True)
 
 
 logger = logging.getLogger(__file__)