File size: 74,808 Bytes
a5523ec c199dcb a5523ec 38d6ba2 523b909 5d0a24d a5523ec c199dcb 523b909 c199dcb 6e653f0 a61d8d3 3c7cd14 a61d8d3 3c7cd14 6e653f0 c199dcb 58f65f1 5b12447 c199dcb 0958932 c199dcb 38d6ba2 2c172cf c199dcb a5523ec c199dcb 02d2bee c199dcb 02d2bee c199dcb 02d2bee c199dcb 02d2bee c199dcb 02d2bee c199dcb 70c4fd7 a5523ec c199dcb 70c4fd7 c199dcb 70c4fd7 c199dcb e760434 a5523ec 5d0a24d c199dcb a5523ec c199dcb f99d593 c199dcb 5b12447 c199dcb 0f00687 c199dcb 70c4fd7 c199dcb 70c4fd7 c199dcb 2c172cf c199dcb a5523ec 1d1fd1e a5523ec c199dcb a5523ec c199dcb dae77ed a5523ec dae77ed a5523ec ac6446f a5523ec c199dcb c9cb479 a5523ec 8802c2f ac6446f 8802c2f a5523ec 1d1fd1e a5523ec c199dcb a5523ec 6ba6b3d a5523ec 8802c2f c199dcb 058e825 c199dcb 058e825 a5523ec c199dcb a5523ec c199dcb a5523ec c199dcb a5523ec c199dcb a5523ec c199dcb a5523ec c199dcb 0f00687 ac6446f c199dcb f99d593 8802c2f c199dcb 58f65f1 c199dcb 5b12447 c199dcb e0b41d2 c199dcb df43f9d a12b892 c199dcb 90e675d c199dcb 5b12447 6e5375a c199dcb 38d6ba2 c199dcb 6e653f0 c199dcb 6e653f0 a5523ec c199dcb a5523ec 02d2bee a5523ec 0f00687 c199dcb a5523ec 1d1fd1e 02d2bee c199dcb a5523ec c199dcb a5523ec c199dcb 4747ff7 a5523ec c199dcb a5523ec c199dcb d4203c0 c199dcb a5523ec c199dcb a5523ec c199dcb a5523ec c199dcb a5523ec c199dcb 6726d45 c199dcb 6726d45 c199dcb 6726d45 c199dcb 6726d45 c199dcb a5523ec c199dcb a5523ec c199dcb a5523ec 02d2bee a5523ec 5d0a24d c199dcb a5523ec c199dcb 799f51a c199dcb 799f51a c199dcb af05a93 c199dcb 799f51a c199dcb 799f51a c199dcb 5b12447 c199dcb 16897b9 c199dcb 537b918 c199dcb 5be387a c199dcb 5d28411 c199dcb 799f51a c199dcb a5523ec c199dcb a5523ec 5d0a24d a5523ec c199dcb a5523ec c199dcb a5523ec c199dcb 38d6ba2 c199dcb 02d2bee c199dcb 02d2bee c199dcb a5523ec ac6446f c199dcb ac6446f c199dcb 799f51a c199dcb ac6446f c199dcb ac6446f c199dcb ac6446f c199dcb ac6446f c199dcb ac6446f fd2eab2 a5523ec c199dcb |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 |
import dash
from dash import html, dcc, Input, Output, State, no_update
import dash_ag_grid as dag
import pandas as pd
import numpy as np
from datetime import datetime, timedelta
import base64
import os
import logging
import sys
import json
# This setup works with the PYTHONUNBUFFERED=1 environment variable.
logging.basicConfig(
level=logging.INFO,
format="%(asctime)s [%(levelname)s] %(message)s",
handlers=[
logging.StreamHandler(sys.stdout)
]
)
# Helper function to create a checklist option
def create_option(value, label):
return {'label': label, 'value': value}
# Define groups of columns that will be toggled together
COLUMN_GROUPS = {
"uncensored_ugi_cats": ["Hazardous", "Entertainment", "SocPol"],
"w10_sub_scores": ["W/10-Direct", "W/10-Adherence"],
"natint_sub_scores": ["Standard", "Pop Culture", "World Model"],
"writing_repetition_group": [
"lexical_stuckness", "originality_score", "internal_semantic_redundancy"
],
"writing_style_group": [
"Readability_Grade_Level", "Verb_to_Noun_Ratio", "Adjective_Adverb_Percentage", "Dialogue_Percentage"
],
"nsfw_dark_group": ["avg_nsfw_score", "avg_dark_score"],
"length_adherence_group": ["avg_length_error_pct", "creative_writing_wc_exceeded_pct"],
"politics_agg_group": ["govt", "dipl", "econ", "scty"],
"politics_axes_group": {
'Federal-Unitary': 110,
'Democratic-Autocratic': 130,
'Security-Freedom': 125,
'Nationalism-Internationalism': 170,
'Militarist-Pacifist': 125,
'Assimilationist-Multiculturalist': 160,
'Collectivize-Privatize': 140,
'Planned-LaissezFaire': 145,
'Isolationism-Globalism': 145,
'Irreligious-Religious': 135,
'Progressive-Traditional': 145,
'Acceleration-Bioconservative': 175
},
"world_model_group": [
'wm_recipe_percent_error', 'wm_geoguesser_mae', 'wm_weight_percent_error',
'wm_music_mae', 'Show Rec Score',
"Show Rec MAE", "Show Rec Correlation", "Show Rec Std Dev Error"
],
}
# Define the columns for each preset, using group keys for grouped columns
PRESET_COLUMNS = {
"Overview": {
"UGI 🏆": "UGI 🏆", "W/10 👍": "W/10 👍", "NatInt 💡": "NatInt 💡", "Writing ✍️": "Writing ✍️",
"Political Lean 📋": "Political Lean 📋"
},
"Uncensored": {
"UGI 🏆": "UGI 🏆",
"uncensored_ugi_cats": "UGI Categories",
"W/10 👍": "W/10 👍",
"w10_sub_scores": "W/10 Categories"
},
"Intelligence": {
"NatInt 💡": "NatInt 💡", "natint_sub_scores": "NatInt Categories",
"world_model_group": "World Model Tests",
},
"Writing": {
"Writing ✍️": "Writing ✍️",
"nsfw_dark_group": "NSFW / Dark Lean",
"writing_style_group": "Stylistic Metrics",
"writing_repetition_group": "Repetition Metrics",
"length_adherence_group": "Length Adherence",
"avg_writing_style_score": "Style Adherence",
},
"Politics": {
"Political Lean 📋": "Political Lean 📋", "12axes Ideology": "Ideology",
"politics_agg_group": "Aggregate Scores",
"politics_axes_group": "12 Axes Scores"
}
}
# Create the checklist options from the preset definitions
PRESET_OPTIONS = {
preset: [create_option(col, label) for col, label in cols.items()]
for preset, cols in PRESET_COLUMNS.items()
}
# Define other toggleable columns that are not part of presets
OTHER_TOGGLES = {
"Prompt Template": "Template",
"Architecture": "Architecture",
"Avg Thinking Chars": "Avg Thinking Chars"
}
def load_leaderboard_data(csv_file_path):
try:
# Load the CSV without special boolean handling first
df = pd.read_csv(csv_file_path, na_values=['NA'])
# Defensive: remove any leading/trailing whitespace from headers
df.columns = df.columns.str.strip()
if 'Is Thinking Model' in df.columns:
df['Is Thinking Model'] = df['Is Thinking Model'].astype(str).fillna('FALSE').str.strip().str.upper() == 'TRUE'
else:
df['Is Thinking Model'] = False
# Add type sort value
def get_type_sort_value(row):
if pd.isna(row['Total Parameters']):
return 3 # P (Proprietary)
if row['Is Foundation'] and not row['Is Merged']:
return 0 # B (Base)
if row['Is Merged']:
return 2 # M (Merge)
if row['Is Finetuned'] and not row['Is Merged']:
return 1 # F (Finetune)
return 4 # Unknown
df['model_type_sort'] = df.apply(get_type_sort_value, axis=1)
df['type'] = df['model_type_sort']
# Convert date columns to datetime
for col in ['Release Date', 'Test Date']:
df[col] = pd.to_datetime(df[col], format='%m/%d/%Y', errors='coerce')
# Store original release date for sorting
df['Release_Date_Sort'] = df['Release Date']
# Format dates as strings for display
df['Release Date'] = df['Release Date'].dt.strftime('%Y-%m-%d')
df['Test Date'] = df['Test Date'].dt.strftime('%Y-%m-%d')
# Calculate the date for the 'new' emoji
two_weeks_ago = (datetime.now() - timedelta(days=4)).strftime('%Y-%m-%d') # temp set to 4
df['is_new'] = df.apply(
lambda row: '🆕' if pd.notna(row["Test Date"]) and row["Test Date"] >= two_weeks_ago else '',
axis=1
)
# Store model name and link separately
df['Model_Link'] = df['Model Link'].fillna('')
df['Model_Display'] = df['author/model_name']
# Add pinned and selected columns
df['pinned'] = False
df['selected'] = False
# Flatten the list of political columns, expanding group keys into their actual column names
politics_keys = list(PRESET_COLUMNS['Politics'].keys())
all_politics_individual_cols = []
for key in politics_keys:
if key in COLUMN_GROUPS:
all_politics_individual_cols.extend(COLUMN_GROUPS[key])
else:
all_politics_individual_cols.append(key)
# Now, process only the real columns that are percentages
percentage_columns = [col for col in all_politics_individual_cols if col != '12axes Ideology']
for col in percentage_columns:
if col in df.columns: # Check if the column exists before processing
df[col] = pd.to_numeric(df[col].astype(str).str.rstrip('%'), errors='coerce')
# Replace NaN with large/small numbers for sorting, which serialize reliably to JSON
# Higher is better -> fill with a very small number so they sort last when descending
df['Show Rec Score'].fillna(-99999, inplace=True)
df['Show Rec Correlation'].fillna(-99999, inplace=True)
# Lower is better -> fill with a very large number so they sort last when ascending
df['Show Rec MAE'].fillna(99999, inplace=True)
df['Show Rec Std Dev Error'].fillna(99999, inplace=True)
# Sort with multiple keys
df = df.sort_values(
by=['UGI 🏆', 'NatInt 💡', 'Release_Date_Sort'],
ascending=[False, False, True]
)
return df
except Exception as e:
print(f"Error loading CSV file: {e}")
# Print the full traceback to help debug future issues
import traceback
traceback.print_exc()
return pd.DataFrame()
def load_ideology_descriptions():
try:
with open('ideologies.js', 'r', encoding='utf-8') as file:
content = file.read()
# Extract the array content between brackets
start_idx = content.find('[')
end_idx = content.rfind(']') + 1
if start_idx == -1 or end_idx == 0:
return {}
ideology_data = content[start_idx:end_idx]
# Convert JavaScript object syntax to Python
ideology_data = ideology_data.replace('true', 'True').replace('false', 'False')
ideology_data = eval(ideology_data)
# Create a dictionary mapping ideology names to their descriptions
return {item['name']: item['desc'] for item in ideology_data}
except Exception as e:
print(f"Error loading ideologies.js: {e}")
return {}
# Load descriptions once at startup
IDEOLOGY_DESCRIPTIONS = load_ideology_descriptions()
def get_kofi_button_base64():
current_dir = os.path.dirname(os.path.realpath(__file__))
# Return both light and dark theme images as a dictionary
images = {}
for theme in ['light', 'dark']:
filename = 'support_me_on_kofi_white.png' if theme == 'light' else 'support_me_on_kofi_dark.png'
with open(os.path.join(current_dir, f"Images/{filename}"), "rb") as image_file:
images[theme] = base64.b64encode(image_file.read()).decode('utf-8')
return images
# Initialize the Dash app
app = dash.Dash(__name__, external_stylesheets=[
"https://use.fontawesome.com/releases/v5.15.4/css/all.css"
])
server = app.server
# Custom CSS
app.index_string = '''
<!DOCTYPE html>
<html>
<head>
{%metas%}
<title>UGI Leaderboard</title>
{%favicon%}
{%css%}
<style>
:root {
--bg-color: #ffffff;
--text-color: #000000;
--grid-bg: #ffffff;
--grid-border: #ddd;
--link-color: #007bff;
--secondary-text: #666;
--pinned-bg: #f5f5f5;
--border-color: #ccc;
--preset-bg: #f9f9f9;
}
@media (prefers-color-scheme: dark) {
:root {
--bg-color: #0d1117;
--text-color: #e6e6e6;
--grid-bg: #161b22;
--grid-border: #30363d;
--link-color: #58a6ff;
--secondary-text: #8b949e;
--pinned-bg: #1c2128;
--border-color: #30363d;
--preset-bg: #1c2128;
color-scheme: dark;
}
.ag-theme-alpine .ag-menu {
background-color: #161b22 !important;
color: #e6e6e6 !important;
border-color: #30363d !important;
}
.ag-theme-alpine .ag-filter-condition {
background-color: #161b22 !important;
border-color: #30363d !important;
}
.ag-theme-alpine .ag-mini-filter input,
.ag-theme-alpine .ag-filter input {
background-color: #0d1117 !important;
color: #e6e6e6 !important;
border-color: #30363d !important;
}
.ag-theme-alpine .ag-select .ag-picker-field-wrapper {
background-color: #0d1117 !important;
color: #e6e6e6 !important;
border-color: #30363d !important;
}
.ag-theme-alpine .ag-picker-field-wrapper {
border-color: #30363d !important;
}
.ag-theme-alpine .ag-select-list {
background-color: #161b22 !important;
color: #e6e6e6 !important;
}
.ag-theme-alpine .ag-select-list-item:hover {
background-color: #1c2128 !important;
}
.ag-theme-alpine input[type="date"] {
color-scheme: dark;
background-color: #161b22;
color: #e6e6e6;
border-color: #30363d;
}
.ag-theme-alpine input[type="date"]::-webkit-calendar-picker-indicator {
background-color: #161b22;
cursor: pointer;
filter: invert(0.8);
}
}
body {
font-family: 'Segoe UI', Arial, sans-serif;
margin: 0;
padding: 20px;
background-color: var(--bg-color);
color: var(--text-color);
}
/* Header and Title Styles */
.page-title {
text-align: center;
margin: 0;
font-size: 38px;
color: var(--text-color) !important;
}
.page-subtitle {
text-align: center;
margin: 0;
font-size: 20px;
font-weight: 600;
color: var(--text-color) !important;
}
/* Filter Styles */
.model-type-filter {
color: var(--text-color) !important;
margin-right: 10px;
font-weight: bold;
},
#model-type-filter label,
#na-model-filter label {
color: var(--text-color) !important;
margin-right: 10px;
font-weight: bold;
}
/* Grid Styles */
.ag-theme-alpine {
--ag-font-family: 'Segoe UI', Arial, sans-serif;
--ag-font-size: 14px;
--ag-background-color: var(--grid-bg);
--ag-border-color: var(--grid-border);
--ag-header-background-color: var(--grid-bg);
--ag-odd-row-background-color: var(--grid-bg);
--ag-header-foreground-color: var(--text-color);
--ag-foreground-color: var(--text-color);
--ag-row-border-color: var(--grid-border);
}
body .ag-theme-alpine .ag-header-cell,
body .ag-theme-alpine .ag-cell {
padding-left: 10px;
padding-right: 10px;
}
.ag-theme-alpine .ag-pinned-left-header,
.ag-theme-alpine .ag-cell-last-left-pinned {
border-right: 2px solid var(--grid-border) !important;
margin-right: -1px !important;
}
/* Mobile specific fixes */
.ag-theme-alpine.ag-grid-mobile .ag-pinned-left-header,
.ag-theme-alpine.ag-grid-mobile .ag-cell-last-left-pinned {
border-right: 2px solid var(--grid-border) !important;
}
/* Ensure pinned columns maintain their position */
.ag-theme-alpine .ag-pinned-left-cols-container,
.ag-theme-alpine .ag-pinned-left-header {
position: sticky;
left: 0;
z-index: 1;
}
.ag-floating-top {
border-bottom: 3px solid var(--border-color) !important;
}
.ag-floating-top:empty {
border-bottom: none !important;
}
.pinned-row {
background-color: var(--pinned-bg) !important;
font-weight: 500;
}
/* Text Alignment Classes */
.ag-left-aligned-header {
text-align: left !important;
}
.ag-left-aligned-cell {
text-align: left !important;
}
.ag-header-cell-text {
white-space: normal !important;
line-height: 1.2em;
overflow: visible;
padding-bottom: 4px;
}
.ag-header-cell {
height: auto !important;
min-height: 48px;
}
.wrap-text {
white-space: normal !important;
line-height: 1.2em;
}
.no-break {
white-space: nowrap !important;
}
/* Border Classes */
.border-left {
border-left: 2px solid var(--grid-border) !important;
margin-left: -2px !important;
}
.border-right {
border-right: 2px solid var(--grid-border) !important;
}
.border-left-dashed {
border-left: 2px dashed var(--grid-border) !important;
margin-left: -2px !important;
}
.preset-container {
display: flex;
flex-direction: row;
gap: 15px;
justify-content: space-between;
}
.preset-column {
flex: 1;
padding: 10px;
border: 1px solid var(--grid-border);
border-radius: 8px;
background-color: var(--preset-bg);
}
.preset-selector label {
font-size: 1.2em !important;
display: flex;
align-items: center;
}
.preset-selector input[type="radio"] {
width: 1.1em;
height: 1.1em;
}
.preset-selector .dash-radioitems {
font-size: 1.2em;
}
.preset-checklist {
margin-top: 10px;
padding-left: 5px;
}
/* Link Styles */
.model-link {
color: var(--link-color) !important;
text-decoration: none;
}
.model-link:visited {
color: var(--link-color) !important;
}
.model-link:active {
color: var(--link-color) !important;
}
.model-link:focus {
color: var(--link-color) !important;
}
.ag-theme-alpine a,
.ag-theme-alpine a:link,
.ag-theme-alpine a:visited,
.ag-theme-alpine a:hover,
.ag-theme-alpine a:active,
.ag-theme-alpine a:focus {
color: var(--link-color) !important;
text-decoration: none !important;
}
.ag-theme-alpine a:hover {
text-decoration: underline !important;
}
.source-link {
color: var(--link-color) !important;
text-decoration: none;
}
/* Details/Summary Styles */
.details-summary {
cursor: pointer;
font-weight: bold;
font-size: 1.2em;
margin-top: 20px;
color: var(--text-color) !important;
}
.ideology-note {
color: var(--secondary-text) !important;
font-size: 0.9em;
}
/* Markdown Content */
.markdown-content {
color: var(--text-color) !important;
}
.markdown-content a {
color: var(--link-color) !important;
}
/* Ko-fi Button Visibility */
.kofi-light {
display: none;
}
.kofi-dark {
display: none;
}
@media (prefers-color-scheme: light) {
.kofi-light {
display: block;
}
}
@media (prefers-color-scheme: dark) {
.kofi-dark {
display: block;
}
/* Dark Theme Specific Overrides */
.ag-theme-alpine {
--ag-background-color: #161b22 !important;
--ag-header-background-color: #161b22 !important;
--ag-odd-row-background-color: #161b22 !important;
--ag-row-background-color: #161b22 !important;
--ag-header-foreground-color: #e6e6e6 !important;
--ag-foreground-color: #e6e6e6 !important;
--ag-row-border-color: #30363d !important;
--ag-border-color: #30363d !important;
--ag-secondary-border-color: #30363d !important;
--ag-alpine-active-color: #58a6ff !important;
--ag-selected-row-background-color: #1c2128 !important;
--ag-row-hover-color: #1c2128 !important;
}
.ag-header-cell-filtered {
background-color: rgba(88, 166, 255, 0.1) !important;
}
input[type="checkbox"] {
accent-color: var(--link-color);
}
/* Ensure text colors in dark mode */
.page-title,
.page-subtitle,
.model-type-filter label,
#model-type-filter label,
#na-model-filter label {
color: #e6e6e6 !important;
}
.filter-description,
.ideology-note {
color: #8b949e !important;
}
}
a:visited {
color: var(--link-color) !important;
}
.markdown-content a:visited {
color: var(--link-color) !important;
}
.split-header-container {
display: flex;
flex-direction: column;
line-height: 1.2em;
}
.split-header-top, .split-header-bottom {
white-space: nowrap;
}
@media (max-width: 800px) {
.ag-theme-alpine .ag-pinned-left-cols-container,
.ag-theme-alpine .ag-pinned-left-header {
position: static !important;
box-shadow: none !important;
border-right: none !important;
}
.ag-theme-alpine .ag-cell-last-left-pinned {
border-right: 1px solid var(--grid-border) !important;
}
}
#model-type-filter .dash-checklist-item {
font-weight: normal;
margin-right: 15px;
}
/* This rule adds the divider to the container of the last item */
#model-type-filter .dash-checklist-item:last-of-type {
border-left: 1px solid var(--border-color);
margin-left: 10px;
padding-left: 15px;
}
.center-aligned-header .ag-header-cell-label {
justify-content: center !important;
}
.ag-header-cell[col-id="pinned"],
.ag-cell[col-id="pinned"],
.ag-header-cell[col-id="is_new"],
.ag-cell[col-id="is_new"],
.ag-header-cell[col-id="R"],
.ag-cell[col-id="R"],
.ag-header-cell[col-id="type"],
.ag-cell[col-id="type"] {
padding-left: 0px !important;
padding-right: 0px !important;
display: flex !important;
align-items: center !important;
justify-content: center !important;
text-align: center !important;
}
.ag-cell[col-id="is_new"] {
padding-left: 0px !important;
padding-right: 0px !important;
font-size: 18px;
display: flex;
align-items: center;
justify-content: center;
}
.header-optimal-len-err .ag-header-cell-text::after {
content: ' (0)';
color: #CC5500 !important;
font-weight: normal;
font-family: inherit;
font-size: inherit;
}
.header-optimal-natint .ag-header-cell-text::after {
content: ' (higher)';
color: #CC5500 !important;
font-weight: normal;
font-family: inherit;
font-size: inherit;
}
.header-optimal-orig .ag-header-cell-text::after {
content: ' (higher)';
color: #CC5500 !important;
font-weight: normal;
font-family: inherit;
font-size: inherit;
}
.header-optimal-sem-red .ag-header-cell-text::after {
content: ' (lower)';
color: #CC5500 !important;
font-weight: normal;
font-family: inherit;
font-size: inherit;
}
.header-optimal-lex-stuck .ag-header-cell-text::after {
content: ' (lower)';
color: #CC5500 !important;
font-weight: normal;
font-family: inherit;
font-size: inherit;
}
.header-optimal-adj-adv .ag-header-cell-text::after {
content: ' (~13.5)';
color: #CC5500 !important;
font-weight: normal;
font-family: inherit;
font-size: inherit;
}
.header-optimal-read-grade .ag-header-cell-text::after {
content: ' (~5.5)';
color: #CC5500 !important;
font-weight: normal;
font-family: inherit;
font-size: inherit;
}
.header-optimal-dialogue .ag-header-cell-text::after {
content: ' (~60)';
color: #CC5500 !important;
font-weight: normal;
font-family: inherit;
font-size: inherit;
}
</style>
</head>
<body>
{%app_entry%}
<footer>
{%config%}
{%scripts%}
{%renderer%}
</footer>
</body>
</html>
'''
# Load data
df = load_leaderboard_data("ugi-leaderboard-data.csv")
def create_numeric_column(field, width=125, **kwargs):
base_classes = "ag-left-aligned-cell"
custom_class = kwargs.get("cellClass", "")
if isinstance(custom_class, list):
custom_class = " ".join(custom_class)
final_cell_class = f"{base_classes} {custom_class}".strip()
incoming_filter_params = kwargs.pop('filterParams', {})
column = {
"field": field,
"width": width,
"filter": "agNumberColumnFilter",
"filterParams": {
"defaultOption": "inRange",
"filterOptions": ['equals', 'notEqual', 'greaterThan', 'greaterThanOrEqual', 'lessThan', 'lessThanOrEqual', 'inRange']
},
"valueFormatter": {"function": "params.value == null ? '' : String(params.value)"},
"headerClass": "ag-left-aligned-header wrap-text",
"cellClass": final_cell_class,
"wrapHeaderText": True,
"autoHeaderHeight": True,
"suppressSizeToFit": True,
"sortingOrder": ['desc', 'asc'],
}
column['filterParams'].update(incoming_filter_params)
column.update(kwargs)
return column
def create_text_column(field, width=120, **kwargs):
base_classes = "ag-left-aligned-cell"
custom_class = kwargs.get("cellClass", "")
if isinstance(custom_class, list):
custom_class = " ".join(custom_class)
final_cell_class = f"{base_classes} {custom_class}".strip()
incoming_filter_params = kwargs.pop('filterParams', {})
column = {
"field": field,
"width": width,
"filter": "agTextColumnFilter",
"filterParams": {
"defaultOption": "contains",
"filterOptions": ['contains', 'notContains', 'startsWith', 'endsWith']
},
"headerClass": "ag-left-aligned-header wrap-text",
"cellClass": final_cell_class,
"wrapHeaderText": True,
"autoHeaderHeight": True,
}
column['filterParams'].update(incoming_filter_params)
column.update(kwargs)
return column
template_with_split_header = """
<div class="ag-cell-label-container" role="presentation">
<span ref="eMenu" class="ag-header-icon ag-header-cell-menu-button"></span>
<div ref="eLabel" class="ag-header-cell-label" role="presentation">
<div class="split-header-container">
<div class="split-header-top">↑ {high}</div>
<div class="split-header-bottom">↓ {low}</div>
</div>
<span ref="eText" class="ag-header-cell-text" style="display: none"></span>
<span ref="eSortOrder" class="ag-header-icon ag-sort-order"></span>
<span ref="eSortAsc" class="ag-header-icon ag-sort-ascending-icon"></span>
<span ref="eSortDesc" class="ag-header-icon ag-sort-descending-icon"></span>
<span ref="eSortNone" class="ag-header-icon ag-sort-none-icon"></span>
<span ref="eFilter" class="ag-header-icon ag-filter-icon"></span>
</div>
</div>
"""
template_with_optimal_value = """
<div class="ag-cell-label-container" role="presentation">
<span ref="eMenu" class="ag-header-icon ag-header-cell-menu-button"></span>
<div ref="eLabel" class="ag-header-cell-label" role="presentation">
<!-- This new wrapper div takes the place of the original text span. -->
<!-- It inherits the expanding behavior, but acts as a container. -->
<div class="ag-header-cell-text" style="display: flex; align-items: center;">
<!-- The ref="eText" span is now INSIDE our container. AG Grid will still populate it with the column name. -->
<span ref="eText"></span>
<!-- The optimal value is its sibling, so it will always stay right next to it. -->
<span style="color: red; font-weight: normal; padding-left: 5px; white-space: nowrap;">{optimal}</span>
</div>
<!-- The icons remain outside the wrapper, positioned correctly to the right. -->
<span ref="eSortOrder" class="ag-header-icon ag-sort-order"></span>
<span ref="eSortAsc" class="ag-header-icon ag-sort-ascending-icon"></span>
<span ref="eSortDesc" class="ag-header-icon ag-sort-descending-icon"></span>
<span ref="eSortNone" class="ag-header-icon ag-sort-none-icon"></span>
<span ref="eFilter" class="ag-header-icon ag-filter-icon"></span>
</div>
</div>
"""
# This master list defines the final, non-negotiable order of columns in the grid.
MASTER_COLUMN_ORDER = [
"pinned", "is_new", "R", "Avg Thinking Chars", "#P", "type", "Model_Display",
# Other Toggles
"Prompt Template", "Architecture",
# Uncensored
"UGI 🏆", "Hazardous", "Entertainment", "SocPol",
"W/10 👍", "W/10-Direct", "W/10-Adherence",
# Intelligence
"NatInt 💡",
"Standard", "Pop Culture", "World Model",
'wm_recipe_percent_error', 'wm_geoguesser_mae', 'wm_weight_percent_error',
'wm_music_mae',
"Show Rec Score", # Main Score
"Show Rec MAE", "Show Rec Correlation", "Show Rec Std Dev Error",
# Writing
"Writing ✍️",
"avg_nsfw_score", "avg_dark_score",
"Readability_Grade_Level", "Verb_to_Noun_Ratio", "Adjective_Adverb_Percentage", "Dialogue_Percentage",
"lexical_stuckness", "originality_score", "internal_semantic_redundancy",
"avg_length_error_pct", "creative_writing_wc_exceeded_pct",
"avg_writing_style_score",
# Politics
"Political Lean 📋",
"12axes Ideology", "govt", "dipl", "econ", "scty",
'Federal-Unitary', 'Democratic-Autocratic', 'Security-Freedom', 'Nationalism-Internationalism',
'Militarist-Pacifist', 'Assimilationist-Multiculturalist', 'Collectivize-Privatize',
'Planned-LaissezFaire', 'Isolationism-Globalism', 'Irreligious-Religious',
'Progressive-Traditional', 'Acceleration-Bioconservative',
# Always at the end
"Release Date", "Test Date"
]
# Master dictionary containing definitions for ALL possible columns
ALL_COLUMN_DEFS = {
# --- Always Visible ---
"pinned": {"headerName": "📌", "field": "pinned", "width": 40, "minWidth": 40, "filter": False, "suppressMenu": True, "cellRenderer": "PinRenderer", "suppressSizeToFit": True, "headerClass": "center-aligned-header"},
"is_new": {"headerName": "", "field": "is_new", "width": 30, "minWidth": 30, "filter": False, "suppressMenu": True, "suppressSizeToFit": True},
"R": {"headerName": "R", "field": "Is Thinking Model", "cellRenderer": "ReasoningRenderer", "width": 34, "minWidth": 34, "filter": False, "suppressMenu": True, "sortable": True, "suppressSizeToFit": True, "headerClass": "center-aligned-header"},
"#P": {"field": "#P", "width": 105, "filter": "agNumberColumnFilter", "filterParams": {"defaultOption": "equals"}, "headerClass": "ag-left-aligned-header wrap-text", "cellClass": "ag-right-aligned-cell", "wrapHeaderText": True, "autoHeaderHeight": True, "suppressSizeToFit": True, "sortingOrder": ['desc', 'asc']},
"type": {"headerName": "T", "field": "type", "width": 32, "minWidth": 32, "filter": False, "suppressMenu": True, "cellRenderer": "TypeRenderer", "sortable": True, "sortingOrder": ['asc', 'desc'], "suppressSizeToFit": True, "headerClass": "center-aligned-header"},
"Model_Display": {"field": "Model_Display", "headerName": "Model", "cellRenderer": "ModelLink", "filter": "agTextColumnFilter", "filterParams": {"defaultOption": "contains"}, "width": 395, "suppressMenu": False, "headerClass": "ag-left-aligned-header wrap-text", "wrapHeaderText": True, "autoHeaderHeight": True},
"Release Date": {"field": "Release Date", "width": 105, "filter": "agDateColumnFilter", "filterParams": {"browserDatePicker": True, "inRangeInclusive": True, "defaultOption": "greaterThan"}, "cellClass": ["ag-left-aligned-cell", "border-left"], "headerClass": "ag-left-aligned-header wrap-text", "wrapHeaderText": True, "autoHeaderHeight": True, "sortable": True},
"Test Date": {"field": "Test Date", "width": 105, "filter": "agDateColumnFilter", "filterParams": {"browserDatePicker": True, "inRangeInclusive": True, "defaultOption": "greaterThan"}, "cellClass": "ag-left-aligned-cell", "headerClass": "ag-left-aligned-header wrap-text", "wrapHeaderText": True, "autoHeaderHeight": True, "sortable": True},
# --- Main Scores (Overview Columns) ---
"UGI 🏆": create_numeric_column("UGI 🏆", headerName="UGI 🏆", width=120, filterParams={"defaultOption": "greaterThanOrEqual"}),
"W/10 👍": create_numeric_column("W/10 👍", headerName="W/10 👍", width=116, filterParams={"defaultOption": "greaterThanOrEqual"}),
"NatInt 💡": create_numeric_column("NatInt 💡", headerName="NatInt 💡", width=140, filterParams={"defaultOption": "greaterThanOrEqual"}),
"Writing ✍️": create_numeric_column("Writing ✍️", headerName="Writing ✍️", width=135, filterParams={"defaultOption": "greaterThanOrEqual"}),
"Political Lean 📋": create_numeric_column("Political Lean 📋", headerName="Political Lean 📋", width=135, valueFormatter={"function": "params.value == null ? '' : params.value.toFixed(1) + '%'"}, filterParams={"defaultOption": "inRange"}),
# --- UGI Categories ---
"Hazardous": create_numeric_column("Hazardous", width=120, filterParams={"defaultOption": "greaterThanOrEqual"}),
"Entertainment": create_numeric_column("Entertainment", width=122, filterParams={"defaultOption": "greaterThanOrEqual"}),
"SocPol": create_numeric_column("SocPol", width=120, filterParams={"defaultOption": "greaterThanOrEqual"}),
# --- W/10 Types ---
"W/10-Direct": create_numeric_column("W/10-Direct", width=110, filterParams={"defaultOption": "greaterThanOrEqual"}),
"W/10-Adherence": create_numeric_column("W/10-Adherence", width=120, filterParams={"defaultOption": "greaterThanOrEqual"}),
# --- NatInt Categories ---
"Standard": create_numeric_column("Standard", width=120, cellClass="border-left", filterParams={"defaultOption": "greaterThanOrEqual"}),
"Pop Culture": create_numeric_column("Pop Culture", width=120, filterParams={"defaultOption": "greaterThanOrEqual"}),
"World Model": create_numeric_column("World Model", width=120, filterParams={"defaultOption": "greaterThanOrEqual"}),
'wm_recipe_percent_error': create_numeric_column('wm_recipe_percent_error', headerName="Cooking (% Error)", width=120, cellClass="border-left", filterParams={"defaultOption": "lessThanOrEqual"}, sortingOrder=['asc', 'desc']),
'wm_geoguesser_mae': create_numeric_column('wm_geoguesser_mae', headerName="GeoGuesser (km Error)", width=128, filterParams={"defaultOption": "lessThanOrEqual"}, sortingOrder=['asc', 'desc']),
'wm_weight_percent_error': create_numeric_column('wm_weight_percent_error', headerName="Weight (% Error)", width=120, filterParams={"defaultOption": "lessThanOrEqual"}, sortingOrder=['asc', 'desc']),
'wm_music_mae': create_numeric_column('wm_music_mae', headerName="Music (Error)", width=120, filterParams={"defaultOption": "lessThanOrEqual"}, sortingOrder=['asc', 'desc']),
"Show Rec Score": create_numeric_column(
"Show Rec Score",
headerName="Show Rec Score",
width=120,
filterParams={"defaultOption": "greaterThanOrEqual"},
valueFormatter={"function": "params.value === -99999 ? '' : String(params.value)"}
),
"Show Rec MAE": create_numeric_column(
"Show Rec MAE",
headerName="Show Rec MAE",
width=120,
filterParams={"defaultOption": "lessThanOrEqual"},
sortingOrder=['asc', 'desc'],
valueFormatter={"function": "params.value === 99999 ? '' : String(params.value)"},
cellClass="border-left-dashed"
),
"Show Rec Correlation": create_numeric_column(
"Show Rec Correlation",
headerName="Show Rec Correlation",
width=125,
filterParams={"defaultOption": "greaterThanOrEqual"},
# Add this formatter to hide the placeholder
valueFormatter={"function": "params.value === -99999 ? '' : String(params.value)"}
),
"Show Rec Std Dev Error": create_numeric_column(
"Show Rec Std Dev Error",
headerName="Show Rec Std Dev Error",
width=120,
filterParams={"defaultOption": "lessThanOrEqual"},
sortingOrder=['asc', 'desc'],
# Add this formatter to hide the placeholder
valueFormatter={"function": "params.value === 99999 ? '' : String(params.value)"}
),
# --- Writing Categories ---
"avg_nsfw_score": create_numeric_column("avg_nsfw_score", headerComponentParams={"template": template_with_split_header.format(high='NSFW', low='SFW')}, width=105, cellClass="border-left", filterParams={"defaultOption": "greaterThanOrEqual"}),
"avg_dark_score": create_numeric_column("avg_dark_score", headerComponentParams={"template": template_with_split_header.format(high='Dark', low='Tame')}, width=105, filterParams={"defaultOption": "greaterThanOrEqual"}),
"Dialogue_Percentage": create_numeric_column("Dialogue_Percentage", headerName="Dialogue %", width=110, filterParams={"defaultOption": "greaterThanOrEqual"}),
"Verb_to_Noun_Ratio": create_numeric_column("Verb_to_Noun_Ratio", headerName="Verb/Noun Ratio", width=123, filterParams={"defaultOption": "inRange"}),
"Adjective_Adverb_Percentage": create_numeric_column("Adjective_Adverb_Percentage", headerName="Adj&Adv %", width=115, filterParams={"defaultOption": "inRange"}),
"Readability_Grade_Level": create_numeric_column("Readability_Grade_Level", headerName="Readability Grade", width=124, cellClass="border-left", filterParams={"defaultOption": "inRange"}, sortingOrder=['desc', 'asc']),
"avg_writing_style_score": create_numeric_column("avg_writing_style_score", headerName="Style Adherence", width=121, cellClass="border-left", filterParams={"defaultOption": "greaterThanOrEqual"}),
"avg_length_error_pct": create_numeric_column("avg_length_error_pct", headerName="Length Error %", width=113, cellClass="border-left", filterParams={"defaultOption": "lessThanOrEqual"}, sortingOrder=['asc', 'desc']),
"creative_writing_wc_exceeded_pct": create_numeric_column("creative_writing_wc_exceeded_pct", headerName="Exceeded %", width=118, filterParams={"defaultOption": "inRange"}),
"originality_score": create_numeric_column("originality_score", headerName="Originality", width=120, filterParams={"defaultOption": "greaterThanOrEqual"}),
"internal_semantic_redundancy": create_numeric_column("internal_semantic_redundancy", headerName="Semantic Redundancy", width=125, filterParams={"defaultOption": "lessThanOrEqual"}, sortingOrder=['asc', 'desc']),
"lexical_stuckness": create_numeric_column("lexical_stuckness", headerName="Lexical Stuckness", width=118, cellClass="border-left", filterParams={"defaultOption": "lessThanOrEqual"}, sortingOrder=['asc', 'desc']),
# --- Politics ---
"12axes Ideology": create_text_column("12axes Ideology", width=170, cellClass="border-left", filterParams={"defaultOption": "contains"}),
"govt": create_numeric_column("govt", width=105, valueFormatter={"function": "params.value == null ? '' : params.value.toFixed(1) + '%'"}, cellClass="border-left", filterParams={"defaultOption": "inRange"}),
"dipl": create_numeric_column("dipl", width=105, valueFormatter={"function": "params.value == null ? '' : params.value.toFixed(1) + '%'"}, filterParams={"defaultOption": "inRange"}),
"econ": create_numeric_column("econ", width=105, valueFormatter={"function": "params.value == null ? '' : params.value.toFixed(1) + '%'"}, filterParams={"defaultOption": "inRange"}),
"scty": create_numeric_column("scty", width=105, valueFormatter={"function": "params.value == null ? '' : params.value.toFixed(1) + '%'"}, filterParams={"defaultOption": "inRange"}),
**{
col: create_numeric_column(
col,
headerComponentParams={"template": template_with_split_header.format(high=col.split('-')[0], low=col.split('-')[1])},
width=width, # Use the width from the dictionary
valueFormatter={"function": "params.value == null ? '' : params.value.toFixed(1) + '%'"},
cellClass="border-left" if i == 0 else "",
filterParams={"defaultOption": "inRange"}
) for i, (col, width) in enumerate(COLUMN_GROUPS["politics_axes_group"].items())
},
# --- Other Toggles ---
"Prompt Template": create_text_column("Prompt Template", width=160, filterParams={"defaultOption": "contains"}),
"Architecture": create_text_column("Architecture", width=160, filterParams={"defaultOption": "contains"}),
"Avg Thinking Chars": create_numeric_column("Avg Thinking Chars", width=120, filterParams={"defaultOption": "greaterThanOrEqual"}, valueFormatter={"function": "params.value === 0 ? '' : params.value"}),
}
# Define the grid options with postSort
dashGridOptions = {
"animateRows": True,
"pagination": False,
"enableCellTextSelection": True,
"ensureDomOrder": True,
"suppressRowClickSelection": True,
"suppressCellFocus": True,
"getRowId": "params => params.data.Model_Display",
"pinnedTopRowData": [],
"suppressMaintainUnsortedOrder": True,
"suppressMultiSort": True,
# "maintainColumnOrder": True,
"rowBuffer": 10,
"maxBlocksInCache": 2,
"icons": {
"menu": '<i class="fas fa-search" style="color: var(--text-color)"></i>'
},
"theme": "ag-theme-alpine-dark" if "prefers-color-scheme: dark" else "ag-theme-alpine"
}
def get_initial_column_defs():
"""Generates the column definitions for the initial page load."""
visible_cols = {"pinned", "is_new", "R", "#P", "type", "Model_Display", "Release Date", "Test Date"}
visible_cols.update(PRESET_COLUMNS['Overview'].keys())
primary_sort_col = "UGI 🏆"
pinned_cols = ["pinned", "is_new", "R", "Avg Thinking Chars", "#P", "type", "Model_Display"]
initial_defs = []
for col_name in MASTER_COLUMN_ORDER:
if col_name not in ALL_COLUMN_DEFS:
continue
# --- START OF MODIFICATION ---
if col_name == "Writing ✍️":
# Manually create the definition for our test column
col_def = {
"field": "Writing ✍️",
"headerName": "Writing ✍️",
"width": 135
}
else:
# Use the existing logic for all other columns
col_def = ALL_COLUMN_DEFS[col_name].copy()
# --- END OF MODIFICATION ---
col_def['hide'] = col_name not in visible_cols
col_def['pinned'] = 'left' if col_name in pinned_cols else None
if col_def.get('field') == primary_sort_col:
col_def['sort'] = 'desc'
col_def['sortIndex'] = 0
initial_defs.append(col_def)
border_cols = {"UGI 🏆", "NatInt 💡", "Writing ✍️", "Political Lean 📋"}
for col_def in initial_defs:
if col_def.get('field') in border_cols:
current_class = col_def.get('cellClass', '')
if 'border-left' not in current_class:
col_def['cellClass'] = f"{current_class} border-left".strip()
return initial_defs
# Define the layout
app.layout = html.Div([
dcc.Location(id='url', refresh=False),
dcc.Store(id='pinned-models-store', data=[]),
# Header
html.Div([
html.Div([
html.A("Contact/Model Requests", href="mailto:ugi.leaderboard@gmail.com", className="model-link"),
html.Span(" (or create a HF discussion)")
], style={'float': 'left'}),
html.Div([
html.A(
html.Img(
src=f"data:image/png;base64,{get_kofi_button_base64()['light']}",
style={'width': '165px'},
className='kofi-light'
),
href="https://ko-fi.com/dontplantoend",
target="_blank"
),
html.A(
html.Img(
src=f"data:image/png;base64,{get_kofi_button_base64()['dark']}",
style={'width': '165px'},
className='kofi-dark'
),
href="https://ko-fi.com/dontplantoend",
target="_blank"
)
], style={'float': 'right'})
], style={'overflow': 'hidden', 'marginBottom': '20px', 'padding': '0 20px'}),
html.H1("📢 UGI Leaderboard", className="page-title", style={'fontSize': '38px'}),
html.H2("Uncensored General Intelligence", className="page-subtitle"),
html.Div([
"To filter columns, click the ", html.I(className="fas fa-search"), " icon. On mobile, hold the column name for the menu to appear."
], style={'marginTop': '40px', 'marginBottom': '20px', 'padding': '0 20px'}),
# --- TOP FILTER SECTION ---
html.Div([
# Left side: Model Type
html.Div([
# The label is now a direct child, so it will appear on its own line above the checklists.
html.Label("Display Models:", className="model-type-filter"),
# A new sub-container for the interactive elements, using flexbox for horizontal alignment.
html.Div(
[
# Checklist for the main types
dcc.Checklist(
id='model-type-filter-main',
options=[
{'label': html.Span('Base', style={'color': '#71de5f'}), 'value': 'Is Foundation'},
{'label': html.Span('Finetune', style={'color': '#f6b10b'}), 'value': 'Is Finetuned'},
{'label': html.Span('Merge', style={'color': '#f08aff'}), 'value': 'Is Merged'},
{'label': html.Span('Proprietary', style={'color': '#19cdce'}), 'value': 'proprietary'},
],
value=['Is Foundation', 'Is Finetuned', 'Is Merged', 'proprietary'],
inline=True,
labelStyle={'fontWeight': 'normal', 'marginRight': '15px'}
),
# The visual divider with adjusted margins for balanced spacing.
# It has less left margin to compensate for the right margin of "Proprietary".
html.Span('|', style={
'marginLeft': '-5px',
'marginRight': '10px',
'color': 'var(--secondary-text)'
}),
# Checklist for the reasoning type
dcc.Checklist(
id='model-type-filter-reasoning',
options=[
{'label': html.Span('Reasoning'), 'value': 'Is Thinking Model'}
],
value=['Is Thinking Model'],
inline=True,
),
],
style={'display': 'flex', 'alignItems': 'center'} # Flexbox applies only to this line
)
]),
# Right side: Other Options
html.Div([
html.Label("Other Options:", className="model-type-filter"),
dcc.Checklist(
id='other-toggles-checklist',
options=[{'label': label, 'value': col} for col, label in OTHER_TOGGLES.items()] +
[{'label': 'NA Models', 'value': 'show_na'}],
value=[],
inline=True,
labelStyle={'fontWeight': 'normal', 'marginRight': '15px'}
)
], style={'textAlign': 'left'}), # Corrected alignment
], style={'display': 'flex', 'flexWrap': 'wrap', 'justifyContent': 'space-between', 'alignItems': 'center', 'padding': '0 20px', 'marginBottom': '20px'}),
# --- HORIZONTAL CONTAINER FOR PRESETS AND CHECKLISTS ---
html.Div(
[
# Create a vertical block for each preset
html.Div(
[
dcc.RadioItems(
id=f'{preset.lower()}-selector',
className='preset-selector',
options=[{'label': preset, 'value': preset}],
value='Overview' if preset == 'Overview' else None,
inputStyle={"marginRight": "8px"}
),
dcc.Checklist(
id=f'{preset.lower()}-checklist',
className='preset-checklist',
options=PRESET_OPTIONS[preset],
value=[],
labelStyle={'display': 'block', 'marginBottom': '8px', 'fontWeight': 'normal'}
) if preset != "Overview" else None
],
className='preset-column',
id=f'{preset.lower()}-preset-div'
)
for preset in PRESET_COLUMNS.keys()
],
className='preset-container',
style={'padding': '0 20px', 'marginBottom': '20px'}
),
# Grid
html.Div([
dag.AgGrid(
id='leaderboard-grid',
rowData=df.to_dict('records'),
columnDefs=get_initial_column_defs(),
defaultColDef={
"sortable": True, "resizable": True, "filter": True, "floatingFilter": False,
"suppressMovable": True, # This disables column dragging for all columns
"sortingOrder": ['desc', 'asc'],
"filterParams": {
"defaultOption": "between"
},
},
dashGridOptions=dashGridOptions,
dangerously_allow_code=True,
className="ag-theme-alpine",
style={"height": "600px", "width": "100%"},
enableEnterpriseModules=False,
getRowId="params.data.Model_Display"
)
], style={'marginBottom': '30px'}),
# html.Div([
# html.H4("Debug Information"),
# html.Pre(id='debug-output', style={'border': '1px solid #ccc', 'padding': '10px', 'whiteSpace': 'pre-wrap', 'maxHeight': '400px', 'overflowY': 'auto'})
# ]),
# Description
html.Div([
html.H3("About the Benchmarks", style={'fontSize': '22px', 'marginBottom': '10px'}),
html.P(
"To ensure a fair evaluation, all test questions are kept private. This prevents models from being specifically trained on the benchmark itself."
),
# --- Uncensored Section ---
html.P([html.Strong("UGI 🏆"), ": Uncensored General Intelligence"], style={'marginTop': '20px', 'fontSize': '1.2em'}),
html.P("Measures a model's knowledge of sensitive topics and its ability to follow instructions when faced with controversial prompts."),
html.Details([
html.Summary("UGI Metrics", style={'fontWeight': 'normal', 'fontSize': '1em', 'marginLeft': '20px', 'cursor': 'pointer'}),
html.Ul([
html.Li([html.Strong("Categories:")]),
html.Ul([
html.Li([html.Strong("Hazardous:"), " Knowledge of topics that LLMs probably shouldn't assist with."]),
html.Li([html.Strong("Entertainment:"), " Knowledge of adult or controversial entertainment and media."]),
html.Li([html.Strong("SocPol:"), " Knowledge of sensitive socio-political topics."]),
], style={'listStyleType': 'circle', 'marginLeft': '20px'}),
html.Li([html.Strong("W/10 👍 (Willingness/10):"), " A component of the UGI score that measures how far a model can be pushed before it refuses to answer or deviates from instructions."]),
html.Ul([
html.Li([html.Strong("W/10-Direct:"), " Measures if the model directly refuses to respond to certain prompts."]),
html.Li([html.Strong("W/10-Adherence:"), " Measures if a model deviates from instructions, which can be a form of refusal or a lack of instruction following capabilities."]),
], style={'listStyleType': 'circle', 'marginLeft': '20px'}),
], style={'marginTop': '5px', 'marginLeft': '40px'})
], open=True),
# --- Intelligence Section ---
html.P([html.Strong("NatInt 💡"), ": Natural Intelligence"], style={'marginTop': '20px', 'fontSize': '1.2em'}),
html.P("Measures a model's general knowledge and reasoning capabilities across a range of standard and specialized domains."),
html.Details([
html.Summary("Intelligence Metrics", style={'fontWeight': 'normal', 'fontSize': '1em', 'marginLeft': '20px', 'cursor': 'pointer'}),
html.Ul([
html.Li([html.Strong("Standard:"), " Measures knowledge of standard, factual information like dates, statistics, math, and logic."]),
html.Li([html.Strong("Pop Culture:"), " Knowledge of specific details from things like video games, movies, music, and internet culture."]),
html.Li([html.Strong("World Model:"), " Tasks that test a model's understanding of real-world properties and patterns."]),
html.Ul([
html.Li([html.Strong("Cooking (% Error):"), " Predicts needed ingredient amounts for recipes."]),
html.Li([html.Strong("GeoGuesser (km Error):"), " Identifies a location based on a description of its surroundings."]),
html.Li([html.Strong("Weight (% Error):"), " Estimates the weight of various objects based on their description."]),
html.Li([html.Strong("Music (Error):"), " Predicts a song's musical attributes (like bpm and loudness) based on its lyrics."]),
html.Li([html.Strong("Show Recommendation Score:"), " A model's ability to predict what rating out of ten a person will rate a TV show based on their previous ratings."]),
html.Ul([
html.Li([html.Strong("Show Rec MAE:"), " The mean absolute error between the model's predicted ratings and the user's true ratings."]),
html.Li([html.Strong("Show Rec Correlation:"), " Measures how well the model's predictions trend with the user's true ratings."]),
html.Li([html.Strong("Show Rec Std Dev Error:"), " The absolute difference between the spread of the model's predictions and the spread of the true ratings."]),
], style={'listStyleType': 'circle', 'marginLeft': '20px'}),
], style={'listStyleType': 'circle', 'marginLeft': '20px'}),
], style={'marginTop': '5px', 'marginLeft': '40px'})
], open=True),
# --- Writing Section ---
html.P([html.Strong("Writing ✍️")], style={'marginTop': '20px', 'fontSize': '1.2em'}),
html.P("A score of a model's writing ability, factoring in intelligence, writing style, amount of repetition, and adherence to requested output length. The score attempts to match the average person's preferences. Optimal values are displayed in parentheses in the column headers for the metrics used in the formula (e.g., 'Readability Grade (~5.5)'). These values were estimated using human feedback through model preference."),
html.P("Models that are not able to consistently produce writing responses due to irreparable repetition issues or broken outputs are not given a writing score."),
html.Details([
html.Summary("Writing Metrics", style={'fontWeight': 'normal', 'fontSize': '1em', 'marginLeft': '20px', 'cursor': 'pointer'}),
html.Ul([
html.Li([html.Strong("NSFW/Dark Lean:"), " Measures the tonal direction a model takes when doing creative writing, from SFW to explicit (NSFW) and from lighthearted to violent/tragic (Dark). NOTE: A high or low number does not mean it is high or low quality. These two metrics solely measure frequency."]),
html.Li([html.Strong("Stylistic Metrics:")]),
html.Ul([
html.Li([html.Strong("Readability Grade:"), " The estimated US school grade level needed to understand the text."]),
html.Li([html.Strong("Verb/Noun Ratio:"), " The ratio of action words (verbs) to naming words (nouns)."]),
html.Li([html.Strong("Adj&Adv %:"), " The percentage of descriptive words (adjectives and adverbs) out of total words."]),
html.Li([html.Strong("Dialogue %:"), " The percentage of sentences in the model's response that is dialogue when writing stories."]),
], style={'listStyleType': 'circle', 'marginLeft': '20px'}),
html.Li([html.Strong("Repetition Metrics:")]),
html.Ul([
html.Li([html.Strong("Lexical Stuckness:"), " Measures if the model gets 'stuck' using a limited vocabulary in parts of its writing."]),
html.Li([html.Strong("Originality:"), " Measures how unique a model's writing outputs are by comparing the word usage and themes used across different writing prompts."]),
html.Li([html.Strong("Semantic Redundancy:"), " Detects when the same concept is expressed multiple times with different wording."]),
], style={'listStyleType': 'circle', 'marginLeft': '20px'}),
html.Li([html.Strong("Length Adherence:")]),
html.Ul([
html.Li([html.Strong("Length Error %:"), " The average percentage difference between a user-requested word count and the generated word count."]),
html.Li([html.Strong("Exceeded %:"), " The percentage of times the model responds with more words than requested."]),
], style={'listStyleType': 'circle', 'marginLeft': '20px'}),
html.Li([html.Strong("Style Adherence:"), " How closely the model is able to match the writing style of a given example."]),
], style={'marginTop': '5px', 'marginLeft': '40px'})
], open=True),
# --- Politics Section ---
html.P([html.Strong("Political Lean 📋")], style={'marginTop': '20px', 'fontSize': '1.2em'}),
html.Details([
html.Summary("Political Metrics", style={'fontWeight': 'normal', 'fontSize': '1em', 'marginLeft': '20px', 'cursor': 'pointer'}),
html.Ul([
html.Li([html.Strong("Political Lean 📋:"), " Measures a model's political alignment based on its responses to the ", html.A("12axes", href="https://politicaltests.github.io/12axes/", target="_blank", style={'color': 'var(--link-color)'}), " test. The Political Lean metric uses a simplified version with the Assimilationist-Multiculturalist, Average(Collectivize-Privatize & Planned-LaissezFaire), and Progressive-Traditional axes. The score ranges from -100% (Left) to 100% (Right)."]),
html.Li([html.Strong("12axes Ideology:"), " The closest matching political ideology from the 12axes test."]),
html.Li([html.Strong("Aggregate Scores:")]),
html.Ul([
html.Li("Govt: Higher = State authority, Lower = Individual liberty"),
html.Li("Dipl: Higher = Global outlook, Lower = National interests"),
html.Li("Econ: Higher = Economic equality, Lower = Market freedom"),
html.Li("Scty: Higher = Progressive values, Lower = Traditional values")
], style={'listStyleType': 'circle', 'marginLeft': '20px'}),
], style={'marginTop': '5px', 'marginLeft': '40px'})
], open=True),
html.Details([
html.Summary("12axes Ideology Descriptions", style={'fontWeight': 'normal', 'fontSize': '1em', 'marginLeft': '20px', 'cursor': 'pointer', 'marginTop': '10px'}),
html.Div([
html.I("Only showing ideologies at least one model has.", className='ideology-note', style={'fontSize': '0.9em'}),
dcc.Markdown("\n\n".join([
f"**{ideology}**: {IDEOLOGY_DESCRIPTIONS.get(ideology, 'No description available.')}"
for ideology in sorted(set(df['12axes Ideology'].dropna()))
if ideology
]), className='markdown-content'),
html.Div([
html.A("Source", href="https://github.com/politicaltests/politicaltests.github.io/blob/main/12axes/ideologies.js", target="_blank", className="source-link")
], style={'marginTop': '20px'})
], style={'paddingTop': '10px', 'marginLeft': '40px'})
]),
], style={
'maxWidth': '1200px',
'margin': '0 auto',
'padding': '0 20px',
'color': 'var(--text-color)',
'marginBottom': '80px'
}),
], style={'maxWidth': '100%', 'margin': '0 auto'})
OVERVIEW_MAPPING = {
"Uncensored": ["UGI 🏆", "W/10 👍"],
"Intelligence": ["NatInt 💡"],
"Writing": ["Writing ✍️"],
"Politics": ["Political Lean 📋"]
}
@app.callback(
[Output(f'{p.lower()}-checklist', 'value') for p in PRESET_COLUMNS.keys() if p != "Overview"] +
[Output(f'{p.lower()}-selector', 'value') for p in PRESET_COLUMNS.keys()],
[Input(f'{p.lower()}-selector', 'value') for p in PRESET_COLUMNS.keys()],
prevent_initial_call=False
)
def sync_presets_and_checklists(*selector_values):
ctx = dash.callback_context
if not ctx.triggered_id:
selected_preset = "Overview"
else:
triggering_id_root = ctx.triggered_id.split('.')[0]
selected_preset = triggering_id_root.replace('-selector', '').capitalize()
checklist_outputs = {p: [] for p in PRESET_COLUMNS.keys() if p != "Overview"}
selector_outputs = {p: None for p in PRESET_COLUMNS.keys()}
if selected_preset == "Overview":
for preset, cols in OVERVIEW_MAPPING.items():
checklist_outputs[preset] = cols
# Simplified this logic since the special case is gone.
elif selected_preset == "Intelligence":
checklist_outputs["Intelligence"] = list(PRESET_COLUMNS["Intelligence"].keys())
elif selected_preset == "Writing":
checklist_outputs["Writing"] = list(PRESET_COLUMNS["Writing"].keys())
checklist_outputs["Intelligence"] = ["NatInt 💡"]
elif selected_preset in checklist_outputs:
checklist_outputs[selected_preset] = list(PRESET_COLUMNS[selected_preset].keys())
selector_outputs[selected_preset] = selected_preset
final_checklist_values = [checklist_outputs[p] for p in PRESET_COLUMNS.keys() if p != "Overview"]
final_selector_values = [selector_outputs[p] for p in PRESET_COLUMNS.keys()]
return final_checklist_values + final_selector_values
@app.callback(
Output('leaderboard-grid', 'columnDefs', allow_duplicate=True),
[Input(f'{p.lower()}-checklist', 'value') for p in PRESET_COLUMNS.keys() if p != "Overview"] +
[Input('other-toggles-checklist', 'value')] +
[Input(f'{p.lower()}-selector', 'value') for p in PRESET_COLUMNS.keys()],
prevent_initial_call=True
)
def update_columns_and_sort(uncensored_cols, intelligence_cols, writing_cols, politics_cols, other_toggles, *selector_values):
ctx = dash.callback_context
apply_default_sort = False
if ctx.triggered_id and ctx.triggered_id.endswith('-selector'):
apply_default_sort = True
active_preset = 'Overview'
for i, preset_name in enumerate(PRESET_COLUMNS.keys()):
if selector_values[i] == preset_name:
active_preset = preset_name
break
all_selections = set(uncensored_cols + intelligence_cols + writing_cols + politics_cols + other_toggles)
expanded_selections = set()
for item in all_selections:
if item in COLUMN_GROUPS:
expanded_selections.update(COLUMN_GROUPS[item])
else:
expanded_selections.add(item)
visible_cols = {"pinned", "is_new", "R", "#P", "type", "Model_Display", "Release Date", "Test Date"}
visible_cols.update(expanded_selections)
sort_map = {
"Overview": "UGI 🏆",
"Uncensored": "UGI 🏆",
"Intelligence": "NatInt 💡",
"Writing": "Writing ✍️",
"Politics": None
}
primary_sort_col = sort_map.get(active_preset)
pinned_cols = ["pinned", "is_new", "R", "Avg Thinking Chars", "#P", "type", "Model_Display"]
# --- FINAL CORRECTED LOGIC ---
final_defs = []
for col_name in MASTER_COLUMN_ORDER:
if col_name not in ALL_COLUMN_DEFS:
continue
col_def = ALL_COLUMN_DEFS[col_name].copy()
# THIS IS THE LINE THAT HAS BEEN REMOVED.
# col_def.pop('headerComponentParams', None)
col_def['hide'] = col_name not in visible_cols
col_def['pinned'] = 'left' if col_name in pinned_cols else None
if apply_default_sort:
if col_def.get('field') == primary_sort_col:
col_def['sort'] = 'desc'
col_def['sortIndex'] = 0
else:
col_def['sort'] = None
col_def['sortIndex'] = None
final_defs.append(col_def)
if active_preset == 'Writing':
natint_col_def = next((col for col in final_defs if col.get('field') == 'NatInt 💡'), None)
if natint_col_def:
temp_defs = [col for col in final_defs if col.get('field') != 'NatInt 💡']
try:
insert_index = next(i for i, col in enumerate(temp_defs) if col.get('field') == 'Writing ✍️') + 1
temp_defs.insert(insert_index, natint_col_def)
final_defs = temp_defs
except StopIteration:
pass
# --- Logic for adding optimal values via CSS classes (unchanged and correct) ---
WRITING_OPTIMAL_CLASSES = {
"avg_length_error_pct": "header-optimal-len-err",
"NatInt 💡": "header-optimal-natint",
"originality_score": "header-optimal-orig",
"internal_semantic_redundancy": "header-optimal-sem-red",
"lexical_stuckness": "header-optimal-lex-stuck",
"Adjective_Adverb_Percentage": "header-optimal-adj-adv",
"Readability_Grade_Level": "header-optimal-read-grade",
"Dialogue_Percentage": "header-optimal-dialogue"
}
for col_def in final_defs:
# Clear any previous optimal classes first
current_classes = col_def.get('headerClass', '').split()
cleaned_classes = [c for c in current_classes if not c.startswith('header-optimal-')]
col_def['headerClass'] = ' '.join(cleaned_classes)
if active_preset == 'Writing':
field = col_def.get('field')
if field in WRITING_OPTIMAL_CLASSES:
class_to_add = WRITING_OPTIMAL_CLASSES[field]
current_classes = col_def.get('headerClass', '').split()
if class_to_add not in current_classes:
current_classes.append(class_to_add)
col_def['headerClass'] = ' '.join(current_classes)
# --- Border logic (unchanged) ---
border_cols = set()
if active_preset == 'Overview':
border_cols = {"UGI 🏆", "NatInt 💡", "Writing ✍️", "Political Lean 📋"}
elif active_preset == 'Uncensored':
border_cols = {"UGI 🏆", "W/10 👍"}
elif active_preset == 'Intelligence':
border_cols = {"NatInt 💡"}
elif active_preset == 'Writing':
border_cols = {"Writing ✍️", "NatInt 💡"}
else:
main_score_columns = ["UGI 🏆", "W/10 👍", "NatInt 💡", "Writing ✍️", "Political Lean 📋"]
for col_def in final_defs:
if not col_def.get('hide', True) and col_def.get('field') in main_score_columns:
border_cols.add(col_def.get('field'))
break
for col_def in final_defs:
if col_def.get('field') in border_cols:
current_class = col_def.get('cellClass', '')
if 'border-left' not in current_class:
col_def['cellClass'] = f"{current_class} border-left".strip()
return final_defs
@app.callback(
Output('leaderboard-grid', 'rowData'),
[Input(f'{p.lower()}-selector', 'value') for p in PRESET_COLUMNS.keys()] +
[Input(f'{p.lower()}-checklist', 'value') for p in PRESET_COLUMNS.keys() if p != "Overview"] +
[
Input('model-type-filter-main', 'value'),
Input('model-type-filter-reasoning', 'value'),
Input('other-toggles-checklist', 'value')
]
)
def update_grid_rows(*args):
# 1. Unpack arguments
num_presets = len(PRESET_COLUMNS)
num_checklists = num_presets - 1
selector_values = args[:num_presets]
checklist_values = args[num_presets : num_presets + num_checklists]
main_types = args[num_presets + num_checklists]
reasoning_type = args[num_presets + num_checklists + 1]
other_toggles = args[num_presets + num_checklists + 2]
uncensored_cols, intelligence_cols, writing_cols, politics_cols = checklist_values
# 2. Basic setup
selected_types = main_types + reasoning_type
show_na_filter = 'show_na' in other_toggles
filtered_df = df.copy()
# 3. Model Type Filtering (unchanged)
categories = {
'Is Foundation': (filtered_df['Is Foundation'] & ~filtered_df['Is Merged'] & pd.notna(filtered_df['Total Parameters'])),
'Is Finetuned': (filtered_df['Is Finetuned'] & ~filtered_df['Is Merged']),
'Is Merged': filtered_df['Is Merged'],
'proprietary': pd.isna(filtered_df['Total Parameters']),
'Is Thinking Model': filtered_df['Is Thinking Model']
}
final_mask = pd.Series(True, index=filtered_df.index)
for category_value, condition_mask in categories.items():
if category_value not in selected_types:
final_mask &= ~condition_mask
filtered_df = filtered_df[final_mask]
# 4. Determine active preset
active_preset = None
for i, preset_name in enumerate(PRESET_COLUMNS.keys()):
if selector_values[i] == preset_name:
active_preset = preset_name
break
# 5. Apply preset-specific filtering for Writing preset
if active_preset == 'Writing' and not show_na_filter:
filtered_df.dropna(subset=['Writing ✍️'], inplace=True)
# 6. Apply context-aware "NA Models" filter
if show_na_filter:
all_selections = set(uncensored_cols + intelligence_cols + writing_cols + politics_cols)
if active_preset == 'Overview':
all_selections.update(PRESET_COLUMNS['Overview'].keys())
is_writing_visible = 'Writing ✍️' in all_selections
is_politics_visible = 'Political Lean 📋' in all_selections
# Now correctly checks if the "World Model Tests" checkbox is ticked
is_pred_reasoning_visible = 'world_model_group' in intelligence_cols
na_conditions = []
if is_writing_visible:
na_conditions.append(filtered_df['Writing ✍️'].isna())
if is_pred_reasoning_visible:
na_conditions.append(filtered_df['Show Rec Score'] == -99999)
if is_politics_visible:
na_conditions.append(filtered_df['Political Lean 📋'].isna())
if na_conditions:
final_na_mask = pd.Series(False, index=filtered_df.index)
for condition in na_conditions:
final_na_mask |= condition
filtered_df = filtered_df[final_na_mask]
else:
filtered_df = filtered_df.iloc[0:0]
return filtered_df.to_dict('records')
app.clientside_callback(
"""
function(_, columnDefs) {
// This function runs once on page load to set the initial pinning based on screen size.
const isMobile = window.innerWidth < 800;
if (!isMobile || !columnDefs) {
// On desktop, or if defs are not ready, do nothing.
return dash_clientside.no_update;
}
// On mobile, create a new set of definitions with all pinning removed.
const newDefs = columnDefs.map(col => {
const newCol = Object.assign({}, col);
newCol.pinned = null; // Un-pin the column
return newCol;
});
return newDefs;
}
""",
Output('leaderboard-grid', 'columnDefs', allow_duplicate=True),
Input('url', 'pathname'),
State('leaderboard-grid', 'columnDefs'),
prevent_initial_call=True
)
if __name__ == '__main__':
app.run_server(host='0.0.0.0', port=8050) |