changing dataset name

app.py

@@ -2665,7 +2665,7 @@ text_to_video_metric_ids = _metric_ids_for(
 datasets = [
     {
         "id": "text_to_video",
-        "name": "Pruna Internal Text-to-Video Benchmark",
+        "name": "VBench-2.0 Dataset",
         "modality": "text_to_video",
         "data": text_to_video_df,
         "columns": text_to_video_display_columns,

ui.py

@@ -80,8 +80,8 @@ and price**. Each view is a **dataset** scored with a **metric**, written as
 
 ## How to read it
 
-1. Pick a **type** (Text to Video, Video to Video, or Text to Image), then a
-   **dataset** and a **metric**.
+1. Pick a **model type** (Text to Video, Video to Video, or Text to Image),
+   then a **dataset** and a **metric**.
 2. **Leaderboards**: ranked by that metric. Price and generation time sit in
    the same table when the source publishes them.
 3. **Pareto plots**: mark models that are not beaten on both higher score
@@ -109,12 +109,12 @@ prompt suites, so samples are not shown.
 
 ## Current datasets
 
-### Pruna Internal Text-to-Video Benchmark
-Pruna's internal text-to-video benchmark, comparing P-Video-2 variants with
-Fal-hosted models. Quality is Datapoint Elo and Rapidata Elo from pairwise
-preference. Price is USD per second of output video. Time per second of
-video is Fal wall time, except Pruna models which use model execution time.
-Samples are not shown for this dataset.
+### VBench-2.0 Dataset
+VBench-2.0 prompts, comparing P-Video-2 variants with Fal-hosted models.
+Quality is Datapoint Elo and Rapidata Elo from pairwise preference. Price
+is USD per second of output video. Time per second of video is Fal wall
+time, except Pruna models which use model execution time. Samples are not
+shown for this dataset.
 
 ### Pruna Internal Video-Edit Benchmark
 Pruna's internal video-to-video editing benchmark, collected by our
@@ -193,7 +193,7 @@ models *within* a Dataset | Metric view.
 - **Prompt counts:** OneIG Alignment uses 100 anime, 100 human, and 99 object
   prompts (299 total). Qwen Image Dataset uses 100 prompts sampled from the
   1,000-prompt pool for roughly even coverage of its fine-grained (L3)
-  categories. The Pruna Internal Text-to-Video Benchmark uses about 90
+  categories. The VBench-2.0 Dataset uses about 90
   generations per model. The Pruna Internal Video-Edit Benchmark uses 78
   prompts across advertising, e-commerce, real estate, camera, lighting,
   text, and related categories. Artificial Analysis and Arena AI use their
@@ -1385,11 +1385,11 @@ def _filter_row(datasets, metrics, default_dataset_id, default_metric_id=None):
         modality_dd = gr.Dropdown(
             choices=_modality_choices(datasets),
             value=default_modality,
-            label="Type",
+            label="Model Type",
             type="value",
             filterable=False,
             scale=1,
-            min_width=150,
+            min_width=170,
         )
         dataset_dd = gr.Dropdown(
             choices=_dataset_choices(datasets, modality=default_modality),
@@ -1439,12 +1439,13 @@ def render_image_workspace(datasets, metrics, default_dataset_id, default_metric
         with gr.Column(elem_classes="workspace-filters") as filters_host:
             gr.Markdown(
                 "<p class='filter-help'>"
-                "Start with Type to switch between Video to Video and Text "
-                "to Image. The rest of the filters follow you across "
-                "Leaderboards, Pareto plots, and Samples. Samples only "
-                "lists datasets and models we have generations for; Pareto "
-                "plots only lists datasets with price or generation time. "
-                "Search in Models, or leave it empty to include every model."
+                "Start with Model Type to switch modalities. The rest of "
+                "the filters follow you across Leaderboards, Pareto plots, "
+                "and Samples. "
+                "Samples only lists datasets and models we have generations "
+                "for; Pareto plots only lists datasets with price or "
+                "generation time. Search in Models, or leave it empty to "
+                "include every model."
                 "</p>",
                 elem_classes="filter-help-host",
             )