fixing id column to hf id

app/app.py

@@ -116,10 +116,10 @@ def _sample_choices(dataset: str, seed: int = 42) -> list:
         rel  = _fmt_gold(row, "relevance_score")
         util = _fmt_gold(row, "utilization_score")
         comp = _fmt_gold(row, "completeness_score")
-        # Use the real HF split + row index when available (loader v2+), else fall back to local pos
+        # Use the real HF id column and split name for cross-referencing in the dataset viewer
         split_name = row.get("source_split", "?")
-        hf_idx     = int(row["source_row_idx"]) if "source_row_idx" in row.index and not pd.isna(row["source_row_idx"]) else i
-        label = f"{truncated} [{split_name}#{hf_idx}]  adh={adh} rel={rel} util={util} comp={comp}"
+        hf_id = int(row["id"]) if "id" in row.index and not pd.isna(row["id"]) else i
+        label = f"{truncated} [{split_name} id={hf_id}]  adh={adh} rel={rel} util={util} comp={comp}"
         choices.append((label, i))
     return choices
 

src/data/ragbench_loader.py

@@ -20,8 +20,7 @@ def load_ragbench_config(config_name: str) -> pd.DataFrame:
     ds = load_dataset("rungalileo/ragbench", config_name)
     df = pd.concat(
         [
-            ds[split].to_pandas()
-            .assign(source_split=split, source_row_idx=range(len(ds[split])))
+            ds[split].to_pandas().assign(source_split=split)
             for split in ds.keys()
         ],
         ignore_index=True,