File size: 12,389 Bytes
7007406 fd38574 b466419 fd38574 de24f31 fd38574 8206722 7007406 b466419 fd38574 fe161c7 b466419 fe161c7 b466419 8206722 b466419 7007406 b466419 8206722 b466419 8206722 b466419 fd38574 77e9502 b466419 77e9502 b466419 77e9502 b466419 77e9502 b466419 8206722 77e9502 fd38574 77e9502 fd38574 b466419 77e9502 b466419 8206722 b466419 8206722 b466419 77e9502 b466419 7007406 b466419 8206722 77e9502 8206722 b466419 8206722 b466419 fd38574 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 |
from dash import Dash, html, dcc, Input, Output
import pandas as pd
from graphs.model_market_share import create_stacked_area_chart, create_world_map, create_range_slider, create_leaderboard
from graphs.model_characteristics import create_concentration_chart, create_line_plot
# Initialize the app
app = Dash()
server = app.server
# Load pre-processed data frames
model_topk_df = pd.read_pickle("data_frames/model_topk_df.pkl")
model_gini_df = pd.read_pickle("data_frames/model_gini_df.pkl")
model_hhi_df = pd.read_pickle("data_frames/model_hhi_df.pkl")
language_concentration_df = pd.read_pickle("data_frames/language_concentration_df.pkl")
license_concentration_df = pd.read_pickle("data_frames/download_license_cumsum_df.pkl")
download_method_cumsum_df = pd.read_pickle("data_frames/download_method_cumsum_df.pkl")
download_arch_cumsum_df = pd.read_pickle("data_frames/download_arch_cumsum_df.pkl")
nat_topk_df = pd.read_pickle("data_frames/nat_topk_df.pkl")
country_concentration_df = pd.read_pickle("data_frames/country_concentration_df.pkl")
author_concentration_df = pd.read_pickle("data_frames/author_concentration_df.pkl")
model_concentration_df = pd.read_pickle("data_frames/model_concentration_df.pkl")
# Configurations
TEMP_MODEL_EVENTS = {
# "Yolo World Mirror": "2024-03-01",
"Llama 3": "2024-04-17",
"Stable Cascade": "2024-02-02",
"Stable Diffusion 3": "2024-05-30",
# "embed/upscale": "2023-03-24",
"DeepSeek-R1": "2025-01-20",
"Gemma-3 12B QAT": "2025-04-15", # gemma-3-12b-it-qat-4bit
# "Qwen": "2025-03-05",
# "Flux RedFlux": "2025-04-12",
# "DeepSeek-V3": "2025-03-24",
# "bloom": "2022-05-19",
"DALLE2-PyTorch": "2022-06-25",
"Stable Diffusion": "2022-08-10",
"CLIP ViT": "2021-01-05",
"YOLOv8": "2023-04-26",
"Sentence Transformer MiniLM v2": "2021-08-30",
}
PALETTE_0 = [
"#335C67",
"#FFF3B0",
"#E09F3E",
"#9E2A2B",
"#540B0E"
]
LANG_SEGMENT_ORDER = [
'Monolingual: EN', 'Monolingual: HR', 'Monolingual: M/LR',
'Multilingual: HR', 'Multilingual', 'Unknown',
]
LICENSE_SEGMENT_ORDER = [
"Open Use", "Open Use (Acceptable Use Policy)", "Open Use (Non-Commercial Only)", "Attribution",
"Acceptable Use Policy", "Non-Commercial Only", "Undocumented", "Undocumented (Acceptable Use Policy)",
]
METHOD_PLOT_CHOICES = {
"cumulative": "none", # none, mean, sum
"y_col": "percent", # percent count
"y_log": False, # True, False
"period": "W",
}
ARCHITECTURE_PLOT_CHOICES = {
"cumulative": "none", # none, mean, sum
"y_col": "percent", # percent count
"y_log": False, # True, False
"period": "W",
}
# Create initial figures
# Model Market Share Tab
model_market_share_area = create_stacked_area_chart(
model_topk_df, model_gini_df, model_hhi_df, TEMP_MODEL_EVENTS, PALETTE_0
)
world_map = create_world_map(
country_concentration_df, "time", "metric", "value"
)
leaderboard = create_leaderboard(
country_concentration_df, author_concentration_df, model_concentration_df
)
slider = create_range_slider(
model_topk_df
)
# Model Characteristics Tab
language_concentration_area = create_concentration_chart(
language_concentration_df, 'time', 'metric', 'value', LANG_SEGMENT_ORDER, PALETTE_0
)
license_concentration_area = create_concentration_chart(
license_concentration_df, 'period', 'status', 'percent', LICENSE_SEGMENT_ORDER, PALETTE_0
)
download_method_cumsum_line = create_line_plot(
download_method_cumsum_df, METHOD_PLOT_CHOICES, PALETTE_0
)
download_arch_cumsum_line = create_line_plot(
download_arch_cumsum_df, ARCHITECTURE_PLOT_CHOICES, PALETTE_0
)
# App layout
app.layout = html.Div(
[
html.Div(
[
html.Div(children='Visualizing the Open Model Ecosystem', style={'fontSize': 28, 'fontWeight': 'bold', 'marginBottom': 6}),
html.Div(children='An interactive dashboard to explore trends in open models on Hugging Face', style={'fontSize': 16, 'marginBottom': 12}),
html.Div(
children=[
html.A(
"Data Provenance Initiative",
href="https://www.dataprovenance.org/",
target="_blank",
style={
'display': 'inline-block',
'padding': '4px 14px',
'fontSize': 13,
'color': 'white',
'backgroundColor': '#2563eb',
'border': 'none',
'borderRadius': '18px',
'textDecoration': 'none',
'fontWeight': 'bold',
'boxShadow': '0 2px 8px rgba(37,99,235,0.08)',
'marginLeft': '6px',
'marginBottom': '4px',
'transition': 'background 0.2s',
'cursor': 'pointer'
}
)
],
style={'fontSize': 14, 'marginBottom': 12}
),
html.Hr(style={'marginTop': 8, 'marginBottom': 8}),
html.Div(children='Lorem Ipsum is simply dummy text of the printing and typesetting industry. Lorem Ipsum has been the industry\'s standard dummy text ever since the 1500s, when an unknown printer took a galley of type and scrambled it to make a type specimen book. It has survived not only five centuries, but also the leap into electronic typesetting, remaining essentially unchanged. It was popularised in the 1960s with the release of Letraset sheets containing Lorem Ipsum passages, and more recently with desktop publishing software like Aldus PageMaker including versions of Lorem Ipsum.', style={'fontSize': 14, 'marginBottom': 12, 'marginLeft': 100, 'marginRight': 100}),
],
style={'textAlign': 'center'}
),
html.Div(
[
dcc.Tabs([
dcc.Tab(label='Model Market Share', children=[
html.Div([
html.Div(children='Select time range to update all graphs below:', style={'fontSize': 16, 'marginBottom': 6, 'marginTop': 10}),
dcc.Graph(figure=slider, id='time-slider', style={'height': '100px'}),
html.Div(
id='output-container-range-slider',
style={
'textAlign': 'center',
'fontSize': 20,
'marginBottom': 15,
'marginTop': 30,
'backgroundColor': 'white',
'borderRadius': '12px',
'boxShadow': '0 2px 12px rgba(0,0,0,0.10)',
'padding': '18px',
'display': 'inline-block',
}
),
], style={'marginBottom': 12, 'justifyContent': 'center', 'textAlign': 'center'}),
html.Div([
dcc.Graph(id='stacked-area-chart'),
], style={'marginBottom': 12}),
html.Div([
html.Div(
dcc.Graph(id='world-map-with-slider'),
style={'display': 'flex', 'justifyContent': 'center'}
),
dcc.Graph(id='leaderboard'),
], style={'marginBottom': 12})
]),
dcc.Tab(label='Model Characteristics', children=[
dcc.Graph(id='language-concentration-chart'),
html.Div([
dcc.Dropdown(['Language Concentration', 'Architecture', 'License', 'Method'], 'Language Concentration', id='dropdown'),
], style={'marginTop': 6}),
]),
dcc.Tab(label='Model Relationships', children=[
]),
])
],
style={
'backgroundColor': 'white',
'borderRadius': '18px',
'boxShadow': '0 4px 24px rgba(0,0,0,0.10)',
'padding': '32px',
'margin': '32px auto',
'maxWidth': '1250px',
}
)
],
style={'fontFamily': 'Inter', 'backgroundColor': '#f7f7fa', 'minHeight': '100vh'}
)
# Callbacks for interactivity
# Model Market Share Tab
# On slider change, update output text
@app.callback(
Output('output-container-range-slider', 'children'),
[Input('time-slider', 'relayoutData')]
)
def update_output(relayout_data):
if relayout_data and 'xaxis.range[0]' in relayout_data and 'xaxis.range[1]' in relayout_data:
start_time = pd.to_datetime(relayout_data['xaxis.range[0]']).strftime('%Y-%m-%d')
end_time = pd.to_datetime(relayout_data['xaxis.range[1]']).strftime('%Y-%m-%d')
return f'Selected time range: {start_time} to {end_time}'
else:
return 'Selected time range: All data'
# On slider change, update world map
@app.callback(
Output('world-map-with-slider', 'figure'),
[Input('time-slider', 'relayoutData')]
)
def update_map(relayout_data):
if relayout_data and 'xaxis.range[0]' in relayout_data and 'xaxis.range[1]' in relayout_data:
start_time = pd.to_datetime(relayout_data['xaxis.range[0]']).strftime('%Y-%m-%d')
end_time = pd.to_datetime(relayout_data['xaxis.range[1]']).strftime('%Y-%m-%d')
updated_fig = create_world_map(
country_concentration_df, "time", "metric", "value", start_time=start_time, end_time=end_time
)
updated_fig.update_layout(font_family="Inter")
return updated_fig
else:
return world_map
# On slider change, update leaderboard
@app.callback(
Output('leaderboard', 'figure'),
[Input('time-slider', 'relayoutData')]
)
def update_leaderboard(relayout_data):
if relayout_data and 'xaxis.range[0]' in relayout_data and 'xaxis.range[1]' in relayout_data:
start_time = pd.to_datetime(relayout_data['xaxis.range[0]']).strftime('%Y-%m-%d')
end_time = pd.to_datetime(relayout_data['xaxis.range[1]']).strftime('%Y-%m-%d')
updated_fig = create_leaderboard(
country_concentration_df, author_concentration_df, model_concentration_df, start_time=start_time, end_time=end_time
)
updated_fig.update_layout(font_family="Inter")
return updated_fig
else:
return leaderboard
# On slider change, update stacked area chart
@app.callback(
Output('stacked-area-chart', 'figure'),
[Input('time-slider', 'relayoutData')]
)
def update_stacked_area(relayout_data):
if relayout_data and 'xaxis.range[0]' in relayout_data and 'xaxis.range[1]' in relayout_data:
start_time = pd.to_datetime(relayout_data['xaxis.range[0]']).strftime('%Y-%m-%d')
end_time = pd.to_datetime(relayout_data['xaxis.range[1]']).strftime('%Y-%m-%d')
updated_fig = create_stacked_area_chart(
model_topk_df, model_gini_df, model_hhi_df, TEMP_MODEL_EVENTS, PALETTE_0,
start_time=start_time, end_time=end_time
)
updated_fig.update_layout(font_family="Inter")
return updated_fig
else:
return model_market_share_area
# Model Characteristics Tab
# On dropdown change, update graph
@app.callback(
Output('language-concentration-chart', 'figure'),
[Input('dropdown', 'value')]
)
def update_graph(selected_metric):
if selected_metric == 'Language Concentration':
return language_concentration_area
elif selected_metric == 'License':
return license_concentration_area
elif selected_metric == 'Method':
return download_method_cumsum_line
elif selected_metric == 'Architecture':
return download_arch_cumsum_line
# Run the app
if __name__ == '__main__':
app.run(debug=True)
|