Fengzhe Zhou
commited on
Commit
·
406a0ba
1
Parent(s):
76c0e18
update leaderboard
Browse files- app.py +19 -28
- data/predict-leaderboard.csv +0 -5
- data/predict-leaderboard.json +301 -0
- data/reason-leaderboard.csv +15 -10
app.py
CHANGED
|
@@ -78,19 +78,7 @@ def create_model_link(model_name):
|
|
| 78 |
# Predict Tab Configuration and Utilities
|
| 79 |
# ============================================================================
|
| 80 |
|
| 81 |
-
#
|
| 82 |
-
PREDICT_COLUMN_NAME_MAPPING = {
|
| 83 |
-
'Common+Misc': 'Common Sense',
|
| 84 |
-
'BG Consistency': 'Background Consistency',
|
| 85 |
-
'Motion': 'Motion Smoothness',
|
| 86 |
-
'Aesthetic': 'Aesthetic Quality',
|
| 87 |
-
'I2V BG': 'I2V Background'
|
| 88 |
-
}
|
| 89 |
-
|
| 90 |
-
# Columns to remove from the dataframe
|
| 91 |
-
PREDICT_COLUMNS_TO_REMOVE = ['Avg Score/Video', 'Common', 'Misc']
|
| 92 |
-
|
| 93 |
-
# Desired column order (using renamed column names)
|
| 94 |
PREDICT_COLUMN_ORDER = [
|
| 95 |
'model',
|
| 96 |
'Overall',
|
|
@@ -168,25 +156,28 @@ PREDICT_NEVER_HIDDEN_COLUMNS = ['model', 'Overall']
|
|
| 168 |
# Columns displayed by default (using renamed column names)
|
| 169 |
PREDICT_DEFAULT_DISPLAYED_COLUMNS = PREDICT_NEVER_HIDDEN_COLUMNS + PREDICT_ALL_SELECTED_COLUMNS
|
| 170 |
|
| 171 |
-
def
|
| 172 |
-
"""
|
| 173 |
-
|
| 174 |
-
|
| 175 |
-
# Remove specified columns
|
| 176 |
-
df = df.drop(columns=PREDICT_COLUMNS_TO_REMOVE, errors='ignore')
|
| 177 |
|
| 178 |
-
|
| 179 |
-
|
|
|
|
|
|
|
| 180 |
|
| 181 |
-
#
|
| 182 |
-
|
| 183 |
-
|
|
|
|
|
|
|
|
|
|
| 184 |
|
| 185 |
-
|
| 186 |
-
|
| 187 |
-
df
|
| 188 |
|
| 189 |
# Format numbers to ensure decimal places (1 decimal for numeric columns)
|
|
|
|
| 190 |
for col in df.columns:
|
| 191 |
if col not in ['model', 'params', 'activate_params'] and pd.api.types.is_numeric_dtype(df[col]):
|
| 192 |
df[col] = df[col].apply(lambda x: f"{x:.1f}" if pd.notna(x) else x)
|
|
@@ -600,7 +591,7 @@ with demo:
|
|
| 600 |
with gr.Tabs(elem_classes="tab-buttons") as tabs:
|
| 601 |
with gr.TabItem("🎨 Predict", elem_id="predict-tab", id=0):
|
| 602 |
# Load data for Predict tab
|
| 603 |
-
predict_df =
|
| 604 |
predict_leaderboard = init_predict_leaderboard(predict_df)
|
| 605 |
|
| 606 |
with gr.TabItem("🔄 Transfer", elem_id="transfer-tab", id=1):
|
|
|
|
| 78 |
# Predict Tab Configuration and Utilities
|
| 79 |
# ============================================================================
|
| 80 |
|
| 81 |
+
# Expected column order (the CSV should already have this order)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 82 |
PREDICT_COLUMN_ORDER = [
|
| 83 |
'model',
|
| 84 |
'Overall',
|
|
|
|
| 156 |
# Columns displayed by default (using renamed column names)
|
| 157 |
PREDICT_DEFAULT_DISPLAYED_COLUMNS = PREDICT_NEVER_HIDDEN_COLUMNS + PREDICT_ALL_SELECTED_COLUMNS
|
| 158 |
|
| 159 |
+
def load_predict_json(json_path):
|
| 160 |
+
"""
|
| 161 |
+
Load predict leaderboard JSON.
|
|
|
|
|
|
|
|
|
|
| 162 |
|
| 163 |
+
The JSON should already be pre-processed by generate_predict_leaderboard.py
|
| 164 |
+
with correct column names, ordering, sorting, and separate model/url fields.
|
| 165 |
+
"""
|
| 166 |
+
df = pd.read_json(json_path, orient='records')
|
| 167 |
|
| 168 |
+
# Convert model name + url to markdown link format for display
|
| 169 |
+
if 'model' in df.columns and 'url' in df.columns:
|
| 170 |
+
def create_link(row):
|
| 171 |
+
if pd.notna(row['url']):
|
| 172 |
+
return f"[{row['model']}]({row['url']})"
|
| 173 |
+
return row['model']
|
| 174 |
|
| 175 |
+
df['model'] = df.apply(create_link, axis=1)
|
| 176 |
+
# Remove the url column as it's now embedded in model
|
| 177 |
+
df = df.drop(columns=['url'])
|
| 178 |
|
| 179 |
# Format numbers to ensure decimal places (1 decimal for numeric columns)
|
| 180 |
+
# Numbers should already be scaled to 0-100 by the generation script
|
| 181 |
for col in df.columns:
|
| 182 |
if col not in ['model', 'params', 'activate_params'] and pd.api.types.is_numeric_dtype(df[col]):
|
| 183 |
df[col] = df[col].apply(lambda x: f"{x:.1f}" if pd.notna(x) else x)
|
|
|
|
| 591 |
with gr.Tabs(elem_classes="tab-buttons") as tabs:
|
| 592 |
with gr.TabItem("🎨 Predict", elem_id="predict-tab", id=0):
|
| 593 |
# Load data for Predict tab
|
| 594 |
+
predict_df = load_predict_json("data/predict-leaderboard.json")
|
| 595 |
predict_leaderboard = init_predict_leaderboard(predict_df)
|
| 596 |
|
| 597 |
with gr.TabItem("🔄 Transfer", elem_id="transfer-tab", id=1):
|
data/predict-leaderboard.csv
DELETED
|
@@ -1,5 +0,0 @@
|
|
| 1 |
-
model,params,activate_params,Overall,AV,Common,Human,Industry,Misc,Physics,Robot,Avg Score/Video,Common+Misc,Domain Score,Aesthetic,BG Consistency,Image Quality,Motion,Overall Consistency,Subject Consistency,I2V BG,I2V Subject,Quality Score
|
| 2 |
-
nvidia/Cosmos-Predict2.5-2B,2.0,2.0,81.0,66.1,95.9,81.4,87.8,91.0,93.9,80.8,84.4,94.1,84.0,52.4,94.2,70.8,99.1,20.1,92.5,97.4,96.6,77.9
|
| 3 |
-
Wan-AI/Wan2.2-I2V-A14B,14.0,14.0,80.6,66.3,94.6,82.1,89.2,90.9,91.8,81.7,84.5,93.2,84.1,51.2,93.7,69.6,98.3,20.4,91.6,96.6,96.0,77.2
|
| 4 |
-
Wan-AI/Wan2.2-TI2V-5B,5.0,5.0,80.4,65.2,95.3,83.0,88.4,89.6,91.5,79.3,84.1,93.1,83.4,51.9,93.7,69.9,98.8,20.3,91.8,96.7,95.9,77.4
|
| 5 |
-
Wan-AI/Wan2.1-I2V-14B-720P,14.0,14.0,79.7,66.9,93.7,80.1,89.7,85.5,88.7,80.1,82.9,90.6,82.7,51.5,93.1,70.1,98.1,20.4,90.0,96.0,95.2,76.8
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
data/predict-leaderboard.json
ADDED
|
@@ -0,0 +1,301 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"model":"Veo-3",
|
| 4 |
+
"url":"https:\/\/deepmind.google\/models\/veo",
|
| 5 |
+
"Overall":82.1,
|
| 6 |
+
"Domain Score":86.7,
|
| 7 |
+
"Quality Score":77.6,
|
| 8 |
+
"Common Sense":94.4,
|
| 9 |
+
"AV":68.7,
|
| 10 |
+
"Robot":86.9,
|
| 11 |
+
"Industry":89.7,
|
| 12 |
+
"Human":84.4,
|
| 13 |
+
"Physics":91.6,
|
| 14 |
+
"Subject Consistency":91.4,
|
| 15 |
+
"Background Consistency":93.1,
|
| 16 |
+
"Motion Smoothness":99.2,
|
| 17 |
+
"Aesthetic Quality":51.9,
|
| 18 |
+
"Image Quality":69.8,
|
| 19 |
+
"Overall Consistency":21.7,
|
| 20 |
+
"I2V Subject":97.0,
|
| 21 |
+
"I2V Background":96.9,
|
| 22 |
+
"params":null,
|
| 23 |
+
"activate_params":null
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"model":"nvidia\/Cosmos-Predict2.5-2B",
|
| 27 |
+
"url":"https:\/\/huggingface.co\/nvidia\/Cosmos-Predict2.5-2B",
|
| 28 |
+
"Overall":81.0,
|
| 29 |
+
"Domain Score":84.0,
|
| 30 |
+
"Quality Score":77.9,
|
| 31 |
+
"Common Sense":94.1,
|
| 32 |
+
"AV":66.1,
|
| 33 |
+
"Robot":80.8,
|
| 34 |
+
"Industry":87.8,
|
| 35 |
+
"Human":81.4,
|
| 36 |
+
"Physics":93.9,
|
| 37 |
+
"Subject Consistency":92.5,
|
| 38 |
+
"Background Consistency":94.2,
|
| 39 |
+
"Motion Smoothness":99.1,
|
| 40 |
+
"Aesthetic Quality":52.4,
|
| 41 |
+
"Image Quality":70.8,
|
| 42 |
+
"Overall Consistency":20.1,
|
| 43 |
+
"I2V Subject":96.6,
|
| 44 |
+
"I2V Background":97.4,
|
| 45 |
+
"params":2.0,
|
| 46 |
+
"activate_params":2.0
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"model":"Wan-AI\/Wan2.2-I2V-A14B",
|
| 50 |
+
"url":"https:\/\/huggingface.co\/Wan-AI\/Wan2.2-I2V-A14B",
|
| 51 |
+
"Overall":80.6,
|
| 52 |
+
"Domain Score":84.1,
|
| 53 |
+
"Quality Score":77.2,
|
| 54 |
+
"Common Sense":93.2,
|
| 55 |
+
"AV":66.3,
|
| 56 |
+
"Robot":81.7,
|
| 57 |
+
"Industry":89.2,
|
| 58 |
+
"Human":82.1,
|
| 59 |
+
"Physics":91.8,
|
| 60 |
+
"Subject Consistency":91.6,
|
| 61 |
+
"Background Consistency":93.7,
|
| 62 |
+
"Motion Smoothness":98.3,
|
| 63 |
+
"Aesthetic Quality":51.2,
|
| 64 |
+
"Image Quality":69.6,
|
| 65 |
+
"Overall Consistency":20.4,
|
| 66 |
+
"I2V Subject":96.0,
|
| 67 |
+
"I2V Background":96.6,
|
| 68 |
+
"params":14.0,
|
| 69 |
+
"activate_params":14.0
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"model":"Wan-AI\/Wan2.2-TI2V-5B",
|
| 73 |
+
"url":"https:\/\/huggingface.co\/Wan-AI\/Wan2.2-TI2V-5B",
|
| 74 |
+
"Overall":80.4,
|
| 75 |
+
"Domain Score":83.4,
|
| 76 |
+
"Quality Score":77.4,
|
| 77 |
+
"Common Sense":93.1,
|
| 78 |
+
"AV":65.2,
|
| 79 |
+
"Robot":79.3,
|
| 80 |
+
"Industry":88.4,
|
| 81 |
+
"Human":83.0,
|
| 82 |
+
"Physics":91.5,
|
| 83 |
+
"Subject Consistency":91.8,
|
| 84 |
+
"Background Consistency":93.7,
|
| 85 |
+
"Motion Smoothness":98.8,
|
| 86 |
+
"Aesthetic Quality":51.9,
|
| 87 |
+
"Image Quality":69.9,
|
| 88 |
+
"Overall Consistency":20.3,
|
| 89 |
+
"I2V Subject":95.9,
|
| 90 |
+
"I2V Background":96.7,
|
| 91 |
+
"params":5.0,
|
| 92 |
+
"activate_params":5.0
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"model":"Wan-AI\/Wan2.1-I2V-14B-720P",
|
| 96 |
+
"url":"https:\/\/huggingface.co\/Wan-AI\/Wan2.1-I2V-14B-720P",
|
| 97 |
+
"Overall":79.7,
|
| 98 |
+
"Domain Score":82.7,
|
| 99 |
+
"Quality Score":76.8,
|
| 100 |
+
"Common Sense":90.6,
|
| 101 |
+
"AV":66.9,
|
| 102 |
+
"Robot":80.1,
|
| 103 |
+
"Industry":89.7,
|
| 104 |
+
"Human":80.1,
|
| 105 |
+
"Physics":88.7,
|
| 106 |
+
"Subject Consistency":90.0,
|
| 107 |
+
"Background Consistency":93.1,
|
| 108 |
+
"Motion Smoothness":98.1,
|
| 109 |
+
"Aesthetic Quality":51.5,
|
| 110 |
+
"Image Quality":70.1,
|
| 111 |
+
"Overall Consistency":20.4,
|
| 112 |
+
"I2V Subject":95.2,
|
| 113 |
+
"I2V Background":96.0,
|
| 114 |
+
"params":14.0,
|
| 115 |
+
"activate_params":14.0
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"model":"MAGI\/MAGI-1-24B",
|
| 119 |
+
"url":"https:\/\/huggingface.co\/sand-ai\/MAGI-1",
|
| 120 |
+
"Overall":78.5,
|
| 121 |
+
"Domain Score":80.5,
|
| 122 |
+
"Quality Score":76.5,
|
| 123 |
+
"Common Sense":90.6,
|
| 124 |
+
"AV":61.8,
|
| 125 |
+
"Robot":73.5,
|
| 126 |
+
"Industry":84.1,
|
| 127 |
+
"Human":79.8,
|
| 128 |
+
"Physics":87.7,
|
| 129 |
+
"Subject Consistency":90.0,
|
| 130 |
+
"Background Consistency":92.4,
|
| 131 |
+
"Motion Smoothness":99.0,
|
| 132 |
+
"Aesthetic Quality":50.2,
|
| 133 |
+
"Image Quality":64.2,
|
| 134 |
+
"Overall Consistency":21.4,
|
| 135 |
+
"I2V Subject":96.8,
|
| 136 |
+
"I2V Background":97.9,
|
| 137 |
+
"params":24.0,
|
| 138 |
+
"activate_params":24.0
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"model":"THUDM\/CogVideoX1.5-5B-I2V",
|
| 142 |
+
"url":"https:\/\/huggingface.co\/THUDM\/CogVideoX1.5-5B-I2V",
|
| 143 |
+
"Overall":78.3,
|
| 144 |
+
"Domain Score":80.1,
|
| 145 |
+
"Quality Score":76.6,
|
| 146 |
+
"Common Sense":89.1,
|
| 147 |
+
"AV":59.7,
|
| 148 |
+
"Robot":73.0,
|
| 149 |
+
"Industry":84.4,
|
| 150 |
+
"Human":79.2,
|
| 151 |
+
"Physics":91.8,
|
| 152 |
+
"Subject Consistency":91.6,
|
| 153 |
+
"Background Consistency":93.9,
|
| 154 |
+
"Motion Smoothness":98.5,
|
| 155 |
+
"Aesthetic Quality":50.0,
|
| 156 |
+
"Image Quality":66.5,
|
| 157 |
+
"Overall Consistency":21.2,
|
| 158 |
+
"I2V Subject":95.0,
|
| 159 |
+
"I2V Background":96.1,
|
| 160 |
+
"params":5.0,
|
| 161 |
+
"activate_params":5.0
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"model":"THUDM\/CogVideoX-5B-I2V",
|
| 165 |
+
"url":"https:\/\/huggingface.co\/THUDM\/CogVideoX-5B-I2V",
|
| 166 |
+
"Overall":77.9,
|
| 167 |
+
"Domain Score":79.5,
|
| 168 |
+
"Quality Score":76.3,
|
| 169 |
+
"Common Sense":87.7,
|
| 170 |
+
"AV":58.0,
|
| 171 |
+
"Robot":74.0,
|
| 172 |
+
"Industry":84.4,
|
| 173 |
+
"Human":79.0,
|
| 174 |
+
"Physics":90.2,
|
| 175 |
+
"Subject Consistency":91.4,
|
| 176 |
+
"Background Consistency":93.4,
|
| 177 |
+
"Motion Smoothness":98.0,
|
| 178 |
+
"Aesthetic Quality":51.2,
|
| 179 |
+
"Image Quality":64.6,
|
| 180 |
+
"Overall Consistency":21.3,
|
| 181 |
+
"I2V Subject":94.1,
|
| 182 |
+
"I2V Background":95.9,
|
| 183 |
+
"params":5.0,
|
| 184 |
+
"activate_params":5.0
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"model":"Lightricks\/LTX-Video-13B",
|
| 188 |
+
"url":"https:\/\/huggingface.co\/Lightricks\/LTX-Video",
|
| 189 |
+
"Overall":77.9,
|
| 190 |
+
"Domain Score":78.4,
|
| 191 |
+
"Quality Score":77.4,
|
| 192 |
+
"Common Sense":88.9,
|
| 193 |
+
"AV":55.3,
|
| 194 |
+
"Robot":70.1,
|
| 195 |
+
"Industry":82.7,
|
| 196 |
+
"Human":78.3,
|
| 197 |
+
"Physics":90.1,
|
| 198 |
+
"Subject Consistency":90.6,
|
| 199 |
+
"Background Consistency":93.5,
|
| 200 |
+
"Motion Smoothness":99.0,
|
| 201 |
+
"Aesthetic Quality":53.5,
|
| 202 |
+
"Image Quality":69.5,
|
| 203 |
+
"Overall Consistency":21.4,
|
| 204 |
+
"I2V Subject":95.7,
|
| 205 |
+
"I2V Background":96.0,
|
| 206 |
+
"params":13.0,
|
| 207 |
+
"activate_params":13.0
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"model":"Tencent\/HunyuanVideo-I2V",
|
| 211 |
+
"url":"https:\/\/huggingface.co\/Tencent\/HunyuanVideo-I2V",
|
| 212 |
+
"Overall":77.4,
|
| 213 |
+
"Domain Score":76.8,
|
| 214 |
+
"Quality Score":78.0,
|
| 215 |
+
"Common Sense":87.4,
|
| 216 |
+
"AV":56.3,
|
| 217 |
+
"Robot":67.7,
|
| 218 |
+
"Industry":83.0,
|
| 219 |
+
"Human":75.5,
|
| 220 |
+
"Physics":88.2,
|
| 221 |
+
"Subject Consistency":94.3,
|
| 222 |
+
"Background Consistency":95.3,
|
| 223 |
+
"Motion Smoothness":99.5,
|
| 224 |
+
"Aesthetic Quality":52.1,
|
| 225 |
+
"Image Quality":65.2,
|
| 226 |
+
"Overall Consistency":21.5,
|
| 227 |
+
"I2V Subject":98.6,
|
| 228 |
+
"I2V Background":97.6,
|
| 229 |
+
"params":null,
|
| 230 |
+
"activate_params":null
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"model":"MAGI\/MAGI-1-4.5B",
|
| 234 |
+
"url":"https:\/\/huggingface.co\/sand-ai\/MAGI-1",
|
| 235 |
+
"Overall":76.9,
|
| 236 |
+
"Domain Score":77.4,
|
| 237 |
+
"Quality Score":76.3,
|
| 238 |
+
"Common Sense":87.5,
|
| 239 |
+
"AV":56.3,
|
| 240 |
+
"Robot":71.6,
|
| 241 |
+
"Industry":79.8,
|
| 242 |
+
"Human":76.0,
|
| 243 |
+
"Physics":88.9,
|
| 244 |
+
"Subject Consistency":92.1,
|
| 245 |
+
"Background Consistency":93.3,
|
| 246 |
+
"Motion Smoothness":99.0,
|
| 247 |
+
"Aesthetic Quality":50.4,
|
| 248 |
+
"Image Quality":61.8,
|
| 249 |
+
"Overall Consistency":21.6,
|
| 250 |
+
"I2V Subject":94.5,
|
| 251 |
+
"I2V Background":98.1,
|
| 252 |
+
"params":4.5,
|
| 253 |
+
"activate_params":4.5
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"model":"Lightricks\/LTX-Video-2B",
|
| 257 |
+
"url":"https:\/\/huggingface.co\/Lightricks\/LTX-Video",
|
| 258 |
+
"Overall":76.9,
|
| 259 |
+
"Domain Score":76.6,
|
| 260 |
+
"Quality Score":77.1,
|
| 261 |
+
"Common Sense":87.3,
|
| 262 |
+
"AV":53.6,
|
| 263 |
+
"Robot":67.1,
|
| 264 |
+
"Industry":81.5,
|
| 265 |
+
"Human":77.1,
|
| 266 |
+
"Physics":87.6,
|
| 267 |
+
"Subject Consistency":89.2,
|
| 268 |
+
"Background Consistency":92.7,
|
| 269 |
+
"Motion Smoothness":98.7,
|
| 270 |
+
"Aesthetic Quality":53.2,
|
| 271 |
+
"Image Quality":71.3,
|
| 272 |
+
"Overall Consistency":21.1,
|
| 273 |
+
"I2V Subject":95.0,
|
| 274 |
+
"I2V Background":95.9,
|
| 275 |
+
"params":2.0,
|
| 276 |
+
"activate_params":2.0
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"model":"Doubiiu\/DynamiCrafter_1024",
|
| 280 |
+
"url":"https:\/\/huggingface.co\/Doubiiu\/DynamiCrafter_1024",
|
| 281 |
+
"Overall":69.7,
|
| 282 |
+
"Domain Score":65.6,
|
| 283 |
+
"Quality Score":73.7,
|
| 284 |
+
"Common Sense":75.2,
|
| 285 |
+
"AV":43.4,
|
| 286 |
+
"Robot":55.0,
|
| 287 |
+
"Industry":72.5,
|
| 288 |
+
"Human":64.1,
|
| 289 |
+
"Physics":83.8,
|
| 290 |
+
"Subject Consistency":91.1,
|
| 291 |
+
"Background Consistency":92.5,
|
| 292 |
+
"Motion Smoothness":94.9,
|
| 293 |
+
"Aesthetic Quality":51.5,
|
| 294 |
+
"Image Quality":68.0,
|
| 295 |
+
"Overall Consistency":21.2,
|
| 296 |
+
"I2V Subject":84.5,
|
| 297 |
+
"I2V Background":86.2,
|
| 298 |
+
"params":null,
|
| 299 |
+
"activate_params":null
|
| 300 |
+
}
|
| 301 |
+
]
|
data/reason-leaderboard.csv
CHANGED
|
@@ -1,10 +1,15 @@
|
|
| 1 |
-
model,
|
| 2 |
-
|
| 3 |
-
Qwen/
|
| 4 |
-
|
| 5 |
-
Qwen/Qwen2.5-VL-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
Qwen/Qwen2-VL-
|
| 10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model,Overall,Common Sense,Embodied Reasoning,Space,Time,Physics,BridgeData V2,RoboVQA,RoboFail,Agibot,HoloAssist,AV,params,activate_params
|
| 2 |
+
GPT-5,70.0,72.7,67.4,67.5,72.8,74.3,53.0,90.9,68.0,55.0,73.0,62.0,,
|
| 3 |
+
Qwen/Qwen3-VL-235B-A22B-Instruct,64.8,65.2,64.4,56.2,69.8,62.4,42.0,93.6,71.0,45.0,76.0,56.0,235.0,22.0
|
| 4 |
+
Qwen/Qwen3-VL-30B-A3B-Instruct,60.6,59.9,61.3,52.5,62.1,59.7,36.0,89.1,67.0,43.0,81.0,49.0,30.0,3.0
|
| 5 |
+
Qwen/Qwen2.5-VL-72B-Instruct,56.8,57.9,55.7,56.2,62.8,52.2,35.0,90.9,73.0,35.0,58.0,39.0,72.0,72.0
|
| 6 |
+
OpenGVLab/InternVL3_5-38B,55.8,55.8,55.7,57.5,60.4,49.1,36.0,81.8,67.0,44.0,71.0,32.0,38.0,38.0
|
| 7 |
+
nvidia/Cosmos-Reason1-7B,54.3,50.7,57.9,57.5,53.7,44.2,41.0,91.8,65.0,42.0,57.0,47.0,7.0,7.0
|
| 8 |
+
GPT-4o,53.7,56.3,51.1,55.0,55.0,58.4,40.0,56.4,65.0,37.0,65.0,43.0,,
|
| 9 |
+
Qwen/Qwen2.5-VL-32B-Instruct,51.9,53.8,50.0,50.0,61.1,45.6,32.0,90.0,52.0,34.0,55.0,33.0,32.0,32.0
|
| 10 |
+
OpenGVLab/InternVL3_5-8B,50.5,50.5,50.5,48.8,54.7,45.6,32.0,77.3,66.0,38.0,49.0,38.0,8.0,8.0
|
| 11 |
+
Qwen/Qwen2.5-VL-7B-Instruct,50.3,47.7,53.0,47.5,55.4,37.6,33.0,83.6,62.0,44.0,47.0,45.0,7.0,7.0
|
| 12 |
+
OpenGVLab/InternVL3_5-14B,49.7,50.3,49.0,52.5,52.0,47.3,26.0,80.0,67.0,28.0,54.0,36.0,14.0,14.0
|
| 13 |
+
OpenGVLab/InternVL3_5-30B-A3B,49.5,49.5,49.5,47.5,54.4,43.8,37.0,78.2,60.0,27.0,55.0,37.0,30.0,3.0
|
| 14 |
+
Qwen/Qwen2.5-VL-3B-Instruct,48.1,47.4,48.9,47.5,50.7,42.9,31.0,82.7,63.0,36.0,48.0,29.0,3.0,3.0
|
| 15 |
+
zai-org/GLM-4.5V,45.5,46.0,44.9,46.2,50.7,39.8,26.0,83.6,69.0,25.0,24.0,38.0,,
|