fix: duplicate model names test in pipeline coder & runner (#763)

* add model name duplicate test in pipeline costeer

* fix ci

Author: XianBW

rdagent/components/coder/data_science/pipeline/eval.py

@@ -71,11 +71,14 @@ def evaluate(
                 model_set_in_scores = set(score_df.index)
 
                 # Check model names (index)
+                if not score_df.index.is_unique:
+                    score_check_text += "\n[Error] The score dataframe contains duplicate model names."
+                    score_ret_code = 1
                 if "ensemble" not in model_set_in_scores:
-                    score_check_text += (
-                        f"\n[Error] The score dataframe doesn't contain the ensemble model.\nscore_df is:\n{score_df}"
-                    )
+                    score_check_text += "\n[Error] The score dataframe doesn't contain the ensemble model."
                     score_ret_code = 1
+                if score_ret_code != 0:
+                    score_check_text += f"The score_df is:\n{score_df}"
 
                 # Check metric name (columns)
                 if score_df.columns.tolist() != [self.scen.metric_name]:

rdagent/log/ui/ds_trace.py

@@ -435,8 +435,8 @@ def summarize_data():
                 df.loc[loop, "End Time (UTC+8)"] = state.times[loop][-1].end + timedelta(hours=8)
             if "running" in loop_data and "no_tag" in loop_data["running"]:
                 try:
-                    df.loc[loop, "Running Score (valid)"] = round(
-                        loop_data["running"]["no_tag"].result.loc["ensemble"].iloc[0], 5
+                    df.loc[loop, "Running Score (valid)"] = str(
+                        round(loop_data["running"]["no_tag"].result.loc["ensemble"].iloc[0], 5)
                     )
                 except:
                     df.loc[loop, "Running Score (valid)"] = "❌"

rdagent/scenarios/data_science/dev/runner/eval.py

@@ -67,9 +67,14 @@ def evaluate(
                 # Check model names (index)
                 # in Pipeline task, we only check ensemble in scores.csv
                 if DS_RD_SETTING.coder_on_whole_pipeline:
+                    if not score_df.index.is_unique:
+                        score_check_text += "\n[Error] The score dataframe contains duplicate model names."
+                        score_ret_code = 1
                     if "ensemble" not in model_set_in_scores:
-                        score_check_text += f"\n[Error] The score dataframe doesn't contain the ensemble model.\nscore_df is:\n{score_df}"
+                        score_check_text += "\n[Error] The score dataframe doesn't contain the ensemble model."
                         score_ret_code = 1
+                    if score_ret_code != 0:
+                        score_check_text += f"The score_df is:\n{score_df}"
                 else:
                     if model_set_in_scores != model_set_in_folder.union({"ensemble"}):
                         score_check_text += f"\n[Error] The scores dataframe does not contain the correct model names as index.\ncorrect model names are: {model_set_in_folder.union({'ensemble'})}\nscore_df is:\n{score_df}"