Valentin Buchner commited on
Commit
968c326
·
1 Parent(s): c951815

adjusted column width and sorting

Browse files
Files changed (3) hide show
  1. README.md +3 -1
  2. leaderboard/app.py +4 -1
  3. leaderboard/leaderboard.json +23 -0
README.md CHANGED
@@ -1,5 +1,5 @@
1
  ---
2
- title: Genception Leaderboard
3
  emoji: 🔥
4
  colorFrom: red
5
  colorTo: green
@@ -7,6 +7,8 @@ sdk: gradio
7
  sdk_version: 4.19.2
8
  app_file: leaderboard/app.py
9
  pinned: true
 
 
10
  ---
11
 
12
  # GenCeption: Evaluate Multimodal LLMs with Unlabeled Unimodal Data
 
1
  ---
2
+ title: GenCeption Leaderboard
3
  emoji: 🔥
4
  colorFrom: red
5
  colorTo: green
 
7
  sdk_version: 4.19.2
8
  app_file: leaderboard/app.py
9
  pinned: true
10
+ short_description: The annotation-free MLLM evaluation benchmark.
11
+ license: mit
12
  ---
13
 
14
  # GenCeption: Evaluate Multimodal LLMs with Unlabeled Unimodal Data
leaderboard/app.py CHANGED
@@ -39,6 +39,7 @@ def create_dataframe(data):
39
  rows.append(row)
40
 
41
  df = pd.DataFrame(rows)
 
42
 
43
  for col in df.columns:
44
  if "Mean" in col:
@@ -52,6 +53,7 @@ def update_display(show_all, df):
52
  legend_visibility = gr.update(visible=show_all)
53
  return filtered_df, legend_visibility
54
 
 
55
  update_data()
56
  demo = gr.Blocks()
57
  with demo:
@@ -60,7 +62,8 @@ with demo:
60
  gr.Markdown(INTRO, elem_classes="markdown-text")
61
  gr.Markdown(INTRO2, elem_classes="markdown-text")
62
  show_all_columns = gr.Checkbox(label="Show all datasets", value=True)
63
- data_display = gr.Dataframe(df, datatype="markdown")
 
64
 
65
  legend_accordion = gr.Accordion("Legend:", open=False, visible=True)
66
  with legend_accordion:
 
39
  rows.append(row)
40
 
41
  df = pd.DataFrame(rows)
42
+ df = df.sort_values(by="Mean", ascending=False)
43
 
44
  for col in df.columns:
45
  if "Mean" in col:
 
53
  legend_visibility = gr.update(visible=show_all)
54
  return filtered_df, legend_visibility
55
 
56
+
57
  update_data()
58
  demo = gr.Blocks()
59
  with demo:
 
62
  gr.Markdown(INTRO, elem_classes="markdown-text")
63
  gr.Markdown(INTRO2, elem_classes="markdown-text")
64
  show_all_columns = gr.Checkbox(label="Show all datasets", value=True)
65
+ column_widths = ["125px"] + [None] * (len(df.columns) - 1)
66
+ data_display = gr.Dataframe(df, datatype="markdown", column_widths=column_widths)
67
 
68
  legend_accordion = gr.Accordion("Legend:", open=False, visible=True)
69
  with legend_accordion:
leaderboard/leaderboard.json CHANGED
@@ -1,5 +1,28 @@
1
  {
2
  "models": [
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  {
4
  "name": "ChatGPT-4V",
5
  "url": "https://cdn.openai.com/papers/GPTV_System_Card.pdf",
 
1
  {
2
  "models": [
3
+ {
4
+ "name": "Claude3-Opus",
5
+ "url": "https://www.anthropic.com/news/claude-3-family",
6
+ "scores": {
7
+ "Mean": 0.340,
8
+ "Exist": 0.382,
9
+ "Count": 0.348,
10
+ "Posi": 0.357,
11
+ "Col": 0.385,
12
+ "Post": 0.360,
13
+ "Cel": 0.317,
14
+ "Sce": 0.374,
15
+ "Lan": 0.344,
16
+ "Art": 0.385,
17
+ "Comm": 0.432,
18
+ "VisMean": 0.368,
19
+ "Code": 0.245,
20
+ "Num": 0.229,
21
+ "Tran": 0.236,
22
+ "OCR": 0.362,
23
+ "TextMean": 0.268
24
+ }
25
+ },
26
  {
27
  "name": "ChatGPT-4V",
28
  "url": "https://cdn.openai.com/papers/GPTV_System_Card.pdf",