Spaces:
Running
Running
File size: 293,750 Bytes
87a53d4 f58c274 87a53d4 f58c274 94fcfa5 87a53d4 74c1053 87a53d4 74c1053 87a53d4 5b894a3 94fcfa5 5b894a3 94fcfa5 5b894a3 94fcfa5 5b894a3 94fcfa5 5b894a3 94fcfa5 5b894a3 94fcfa5 5b894a3 94fcfa5 5b894a3 94fcfa5 5b894a3 94fcfa5 5b894a3 94fcfa5 5b894a3 c2dba37 5b894a3 87a53d4 f58c274 87a53d4 632b302 87a53d4 632b302 87a53d4 555b191 632b302 87a53d4 632b302 87a53d4 f4ec8ea 87a53d4 632b302 87a53d4 f4ec8ea c2dba37 267d7c2 87a53d4 94fcfa5 87a53d4 c2dba37 87a53d4 f4ec8ea 87a53d4 7ac2596 87a53d4 94fcfa5 87a53d4 f4ec8ea 87a53d4 5b894a3 f58c274 5b894a3 f4ec8ea 5b894a3 7ac2596 f4ec8ea 227ea73 f4ec8ea 5b894a3 f4ec8ea 5b894a3 7ac2596 5b894a3 227ea73 5b894a3 f4ec8ea 5b894a3 f4ec8ea 5b894a3 7ac2596 5b894a3 227ea73 5b894a3 f4ec8ea 87a53d4 5b894a3 87a53d4 5b894a3 87a53d4 5b894a3 227ea73 5b894a3 227ea73 5b894a3 87a53d4 5b894a3 f58c274 5b894a3 87a53d4 f58c274 5b894a3 87a53d4 5b894a3 f58c274 5b894a3 f4ec8ea 5b894a3 f4ec8ea 5b894a3 227ea73 5b894a3 227ea73 5b894a3 227ea73 5b894a3 db2bd37 5b894a3 227ea73 db2bd37 5b894a3 227ea73 5b894a3 227ea73 5b894a3 87a53d4 f58c274 5b894a3 f58c274 87a53d4 5b894a3 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 5b894a3 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 5b894a3 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 5b894a3 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 74c1053 75d2102 5b894a3 75d2102 5b894a3 75d2102 f4ec8ea 9ed293f bd767d8 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 632b302 9ed293f bd767d8 9ed293f 87a53d4 9ed293f 5b894a3 87a53d4 5b894a3 f58c274 87a53d4 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 f4ec8ea 5b894a3 9ed293f 5b894a3 9ed293f 5b894a3 9ed293f bd767d8 9ed293f bd767d8 9ed293f 5b894a3 9ed293f bd767d8 9ed293f bd767d8 9ed293f bd767d8 9ed293f 5b894a3 87a53d4 9ed293f f58c274 87a53d4 5b894a3 f4ec8ea 77de5bf f4ec8ea 5b894a3 77de5bf c2dba37 77de5bf 7ac2596 94fcfa5 77de5bf 5b894a3 77de5bf 5b894a3 77de5bf 5b894a3 77de5bf 5b894a3 77de5bf 632b302 5b894a3 77de5bf 5b894a3 77de5bf 5b894a3 77de5bf 5b894a3 77de5bf c2dba37 77de5bf 5b894a3 77de5bf 5b894a3 77de5bf c2dba37 77de5bf 5b894a3 f4ec8ea 77de5bf 632b302 77de5bf 632b302 77de5bf 632b302 77de5bf 632b302 77de5bf 632b302 77de5bf 632b302 77de5bf 632b302 77de5bf f4ec8ea 5b894a3 555b191 f58c274 f4ec8ea 87a53d4 5b894a3 87a53d4 5b894a3 632b302 5b894a3 267d7c2 5b894a3 c2dba37 5b894a3 267d7c2 94fcfa5 267d7c2 5b894a3 632b302 5b894a3 87a53d4 5b894a3 87a53d4 f58c274 | 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 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553 4554 4555 4556 4557 4558 4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 | <!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<meta name="description"
content="FLaME: Holistic Financial Language Model Evaluation - A comprehensive benchmarking suite for evaluating language models on financial NLP tasks.">
<meta name="keywords" content="FLaME, Finance, Language Models, NLP, Benchmarking, FinNLP">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>FLaME: Holistic Financial Language Model Evaluation</title>
<link rel="icon" href="./static/images/FLaME_favicon.png" type="image/png">
<link href="https://fonts.googleapis.com/css?family=Google+Sans|Noto+Sans|Castoro"
rel="stylesheet">
<link rel="stylesheet" href="./static/css/bulma.min.css">
<link rel="stylesheet" href="./static/css/bulma-carousel.min.css">
<link rel="stylesheet" href="./static/css/bulma-slider.min.css">
<link rel="stylesheet" href="./static/css/fontawesome.all.min.css">
<link rel="stylesheet"
href="https://cdn.jsdelivr.net/gh/jpswalsh/academicons@1/css/academicons.min.css">
<link rel="stylesheet" href="./static/css/index.css">
<link rel="stylesheet" href="./static/css/results.css">
<link rel="stylesheet" href="./static/css/tooltips.css">
<link rel="icon" href="./static/images/favicon.svg">
<script src="https://ajax.googleapis.com/ajax/libs/jquery/3.5.1/jquery.min.js"></script>
<script defer src="./static/js/fontawesome.all.min.js"></script>
<script src="./static/js/bulma-carousel.min.js"></script>
<script src="./static/js/bulma-slider.min.js"></script>
<script src="./static/js/index.js"></script>
<script src="./static/js/results.js"></script>
<script src="./static/js/tooltips.js"></script>
<script src="./static/js/fixed-tooltips.js"></script>
<script src="./static/js/tooltip-fix.js"></script>
<script src="./static/js/model-tooltips.js"></script>
</head>
<body>
<!-- Navigation bar -->
<nav class="navbar is-fixed-top" role="navigation" aria-label="main navigation" style="background-color: #004d99; box-shadow: 0 2px 10px rgba(0,0,0,0.1);">
<div class="container">
<div class="navbar-brand">
<a class="navbar-item" href="#" style="color: white;">
<img src="./static/images/FLaME_favicon.png" alt="FLaME Logo" style="width: 30px; height: 30px;">
<span class="flame">FLaME</span>
</a>
<a role="button" class="navbar-burger" aria-label="menu" aria-expanded="false" data-target="navbarMenu">
<span aria-hidden="true"></span>
<span aria-hidden="true"></span>
<span aria-hidden="true"></span>
</a>
</div>
<div id="navbarMenu" class="navbar-menu" style="background-color: #004d99;">
<div class="navbar-end">
<a class="navbar-item" href="#abstract" style="color: white;">
Abstract
</a>
<a class="navbar-item" href="#findings" style="color: white;">
Findings
</a>
<a class="navbar-item" href="#methodology" style="color: white;">
Methodology
</a>
<a class="navbar-item" href="#results" style="color: white;">
Results
</a>
<a class="navbar-item" href="#framework" style="color: white;">
Framework
</a>
<a class="navbar-item" href="#BibTeX" style="color: white;">
Citation
</a>
<div class="navbar-item" style="color: white;">
<div class="buttons">
<a class="button is-info" href="https://github.com/gtfintechlab/FLaME" target="_blank">
<span class="icon">
<i class="fab fa-github"></i>
</span>
<span>GitHub</span>
</a>
</div>
</div>
</div>
</div>
</div>
</nav>
<section class="hero has-text-centered pt-6">
<div class="hero-body">
<div class="container is-max-desktop">
<div class="columns is-centered">
<div class="column has-text-centered">
<h1 class="title is-1 publication-title">FLaME: Holistic Financial Language Model Evaluation</h1>
<div class="is-size-5 publication-authors">
<span class="author-block">
<a href="#" target="_blank">Glenn Matlin</a><sup>1</sup>,</span>
<span class="author-block">
<a href="#" target="_blank">Mika Okamoto</a><sup>1</sup>,</span>
<span class="author-block">
<a href="#" target="_blank">Huzaifa Pardawala</a><sup>1</sup>,</span>
<span class="author-block">
<a href="#" target="_blank">Yang Yang</a><sup>1</sup>,</span>
<span class="author-block">
<a href="#" target="_blank">Sudheer Chava</a><sup>1</sup>
</span>
</div>
<div class="is-size-5 publication-authors">
<span class="author-block"><sup>1</sup>Georgia Institute of Technology</span>
</div>
<div class="column has-text-centered">
<div class="publication-links">
<!-- PDF Link. -->
<span class="link-block">
<a href="FLaME/FLaME.pdf" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fas fa-file-pdf"></i>
</span>
<span>Paper</span>
</a>
</span>
<!-- ArXiv Link -->
<!-- <span class="link-block">
<a href="https://arxiv.org/abs/" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="ai ai-arxiv"></i>
</span>
<span>arXiv</span>
</a>
</span> -->
<!-- Code Link. -->
<span class="link-block">
<a href="https://github.com/gtfintechlab/FLaME" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<i class="fab fa-github"></i>
</span>
<span>Code</span>
</a>
</span>
<!-- HuggingFace Link. -->
<span class="link-block">
<a href="https://huggingface.co/gtfintechlab/" target="_blank"
class="external-link button is-normal is-rounded is-dark">
<span class="icon">
<img src="./static/images/huggingface_logo.svg" alt="HuggingFace" width="20" height="20">
</span>
<span>HuggingFace</span>
</a>
</div>
</div>
</div>
</div>
</div>
</div>
</section>
<section class="hero teaser">
<div class="container is-max-desktop">
<div class="hero-body">
<h2 class="subtitle has-text-centered mb-5">
<span class="flame">FLaME</span> is the first comprehensive benchmarking suite for evaluating language models on financial NLP tasks.
</h2>
<div class="has-text-centered">
<!-- Main overview figure -->
<div class="figure-container mb-5">
<figure class="image">
<!-- Note: In production, convert PDF to PNG/JPG -->
<img src="FLaME/content/figures/FLaME-fig_overview_flow.png" alt="FLaME Evaluation Framework">
<figcaption class="has-text-centered mt-2 is-italic">
<strong>Figure 1:</strong> Functional Overview of FLaME. The evaluation framework includes a comprehensive taxonomy, carefully selected datasets, diverse models, and standardized metrics.
</figcaption>
</figure>
</div>
<!-- Framework visualization with cards -->
<div class="columns is-multiline mt-5">
<!-- Task Taxonomy -->
<div class="column is-half">
<div class="card has-background-light h-100">
<div class="card-header">
<p class="card-header-title">
<span class="icon mr-2"><i class="fas fa-tasks"></i></span>
Task Taxonomy
</p>
</div>
<div class="card-content">
<div class="content">
<figure class="image">
<!-- Note: In production, convert PDF to PNG/JPG -->
<img src="FLaME/content/figures/FLaME-fig_methodology_tasks.png" alt="FLaME Task Taxonomy">
</figure>
<p class="mt-3"><strong>Task Taxonomy:</strong> Illustrative breakdown for each of the six core NLP task categories (Classification, Sentiment Analysis, Information Retrieval, Causal Analysis, Summarization, and Question Answering). Each category encompasses specialized variants depending on data format, user needs, and domain constraints.</p>
</div>
</div>
</div>
</div>
<!-- Domain Taxonomy -->
<div class="column is-half">
<div class="card has-background-light h-100">
<div class="card-header">
<p class="card-header-title">
<span class="icon mr-2"><i class="fas fa-building"></i></span>
Domain Taxonomy
</p>
</div>
<div class="card-content">
<div class="content">
<figure class="image">
<!-- Note: In production, convert PDF to PNG/JPG -->
<img src="FLaME/content/figures/FLaME-fig_methodology_domain.png" alt="FLaME Domain Taxonomy">
</figure>
<p class="mt-3"><strong>Domain Taxonomy:</strong> Holistic taxonomy for FLaME. Unlike previous FinNLP benchmarks that were tied to specific tasks with single metrics, FLaME takes a comprehensive approach by mapping the full space of tasks, scenarios, and metrics across multiple dimensions for complete analysis.</p>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</section>
<section id="abstract" class="section content-section">
<div class="container is-max-desktop">
<!-- Abstract. -->
<div class="columns is-centered">
<div class="column is-four-fifths">
<h2 class="title is-3 section-title has-text-centered">Abstract</h2>
<div class="content has-text-justified abstract-content">
<div class="abstract-box p-5 has-background-white-ter">
<p>
Language Models (LMs) have demonstrated impressive capabilities with core Natural Language Processing (NLP) tasks. The effectiveness of LMs for highly specialized knowledge-intensive tasks in finance remains difficult to assess due to major gaps in the methodologies of existing evaluation frameworks, which have caused an erroneous belief in a far lower bound of LMs' performance on common Finance NLP (FinNLP) tasks.
</p>
<p>
To address this issue, we present <span class="flame">FLaME</span> (Financial Language Model Evaluation), the first comprehensive benchmark suite for evaluating language models on financial NLP tasks. <span class="flame">FLaME</span> includes a standardized evaluation framework, carefully selected datasets across six core financial tasks, and extensive testing across both open and proprietary LMs. Our benchmark assesses models on text classification, sentiment analysis, information retrieval, causal analysis, summarization, and question answering within the financial domain.
</p>
<p>
Our results reveal significant variations in model performance across different financial tasks and domains, with no single model emerging as universally superior. We also find that performance does not always correlate with model size, and that open-weight models can offer strong cost-performance efficiency compared to their proprietary counterparts. <span class="flame">FLaME</span> provides a foundation for building, testing, and advancing specialized NLP models for finance.
</p>
</div>
<div class="paper-stats has-text-centered mt-5">
<div class="columns is-mobile is-multiline">
<div class="column">
<div class="stat-box p-3">
<span class="icon is-large"><i class="fas fa-tasks fa-2x"></i></span>
<p class="is-size-5 mt-2"><strong>6</strong></p>
<p>FinNLP Task Categories</p>
</div>
</div>
<div class="column">
<div class="stat-box p-3">
<span class="icon is-large"><i class="fas fa-database fa-2x"></i></span>
<p class="is-size-5 mt-2"><strong>19</strong></p>
<p>Datasets</p>
</div>
</div>
<div class="column">
<div class="stat-box p-3">
<span class="icon is-large"><i class="fas fa-robot fa-2x"></i></span>
<p class="is-size-5 mt-2"><strong>24</strong></p>
<p>LLMs Evaluated</p>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
<!--/ Abstract. -->
</div>
</section>
<section id="findings" class="section content-section">
<div class="container is-max-desktop">
<!-- Key Findings -->
<div class="columns is-centered">
<div class="column is-full">
<h2 class="title is-3 section-title has-text-centered">Key Findings</h2>
<div class="content">
<div class="notification is-light mb-5">
<p class="is-size-5 has-text-centered mb-4">
Our comprehensive evaluation reveals several important insights about language model performance on financial tasks:
</p>
</div>
<!-- Cards for key findings -->
<div class="columns is-multiline">
<!-- Finding 1 -->
<div class="column is-one-third">
<div class="card h-100">
<div class="card-content">
<div class="content">
<span class="icon is-large has-text-primary mb-3">
<i class="fas fa-trophy fa-2x"></i>
</span>
<h4 class="title is-5">No Universal Winner</h4>
<p>No single language model performs best across all financial NLP tasks</p>
</div>
</div>
</div>
</div>
<!-- Finding 2 -->
<div class="column is-one-third">
<div class="card h-100">
<div class="card-content">
<div class="content">
<span class="icon is-large has-text-primary mb-3">
<i class="fas fa-chart-line fa-2x"></i>
</span>
<h4 class="title is-5">Domain Variations</h4>
<p>Performance varies significantly based on domain and task structure</p>
</div>
</div>
</div>
</div>
<!-- Finding 3 -->
<div class="column is-one-third">
<div class="card h-100">
<div class="card-content">
<div class="content">
<span class="icon is-large has-text-primary mb-3">
<i class="fas fa-balance-scale fa-2x"></i>
</span>
<h4 class="title is-5">Cost Efficiency</h4>
<p>Open-weight and mid-scale models show strong cost/performance efficiency</p>
</div>
</div>
</div>
</div>
<!-- Finding 4 -->
<div class="column is-one-third">
<div class="card h-100">
<div class="card-content">
<div class="content">
<span class="icon is-large has-text-primary mb-3">
<i class="fas fa-calculator fa-2x"></i>
</span>
<h4 class="title is-5">Numeric Challenges</h4>
<p>Numeric reasoning tasks (FNXL, ConvFinQA) remain challenging for all models</p>
</div>
</div>
</div>
</div>
<!-- Finding 5 -->
<div class="column is-one-third">
<div class="card h-100">
<div class="card-content">
<div class="content">
<span class="icon is-large has-text-primary mb-3">
<i class="fas fa-ruler fa-2x"></i>
</span>
<h4 class="title is-5">Size Doesn't Matter</h4>
<p>There's often no consistent correlation between model size and performance</p>
</div>
</div>
</div>
</div>
<!-- Finding 6 -->
<div class="column is-one-third">
<div class="card h-100">
<div class="card-content">
<div class="content">
<span class="icon is-large has-text-primary mb-3">
<i class="fas fa-award fa-2x"></i>
</span>
<h4 class="title is-5">Top Performers</h4>
<p>DeepSeek R1, OpenAI o1-mini, and Claude 3.5 Sonnet demonstrate strong capabilities</p>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
<!--/ Key Findings -->
</div>
</section>
<section id="methodology" class="section content-section">
<div class="container is-max-desktop">
<!-- Methodology -->
<div class="columns is-centered">
<div class="column is-full">
<h2 class="title is-3 section-title has-text-centered">Methodology</h2>
<div class="content">
<div class="notification is-primary is-light mb-5">
<p class="is-size-5 has-text-centered mb-0">
Our methodological framework consists of four key components designed to provide comprehensive evaluation of language models on financial tasks.
</p>
</div>
<!-- Methodology workflow diagram -->
<div class="methodology-diagram has-text-centered mb-6">
<div class="columns is-mobile is-multiline">
<div class="column is-3">
<div class="methodology-step">
<div class="step-number">1</div>
<div class="step-icon">
<span class="icon is-large"><i class="fas fa-sitemap fa-2x"></i></span>
</div>
<h5 class="title is-6 mt-2">Taxonomy</h5>
<p class="is-size-7">Define task categories</p>
</div>
</div>
<div class="column is-3">
<div class="methodology-step">
<div class="step-number">2</div>
<div class="step-icon">
<span class="icon is-large"><i class="fas fa-database fa-2x"></i></span>
</div>
<h5 class="title is-6 mt-2">Datasets</h5>
<p class="is-size-7">Select & prepare data</p>
</div>
</div>
<div class="column is-3">
<div class="methodology-step">
<div class="step-number">3</div>
<div class="step-icon">
<span class="icon is-large"><i class="fas fa-robot fa-2x"></i></span>
</div>
<h5 class="title is-6 mt-2">Models</h5>
<p class="is-size-7">Benchmark LMs</p>
</div>
</div>
<div class="column is-3">
<div class="methodology-step">
<div class="step-number">4</div>
<div class="step-icon">
<span class="icon is-large"><i class="fas fa-chart-bar fa-2x"></i></span>
</div>
<h5 class="title is-6 mt-2">Evaluation</h5>
<p class="is-size-7">Measure performance</p>
</div>
</div>
</div>
</div>
<!-- Methodology detailed info -->
<div class="columns is-multiline">
<!-- Taxonomy -->
<div class="column is-6">
<div class="card h-100">
<div class="card-header">
<p class="card-header-title">
<span class="icon mr-2"><i class="fas fa-sitemap"></i></span>
Taxonomy
</p>
</div>
<div class="card-content">
<div class="content">
<p>We developed a scenario-based taxonomy organizing financial NLP tasks along three dimensions:</p>
<ul style="list-style-type: none;">
<li><strong>Tasks:</strong> Six core tasks - text classification, sentiment analysis, information retrieval, causal analysis, text summarization, and question answering</li>
<li><strong>Domains:</strong> Categorized by what (type of data), who (data source), where (origination), when (time period), how (generation method), and why (purpose)</li>
<li><strong>Languages:</strong> Currently focusing on English with identified need for multilingual expansion</li>
</ul>
</div>
</div>
</div>
</div>
<!-- Datasets -->
<div class="column is-6">
<div class="card h-100">
<div class="card-header">
<p class="card-header-title">
<span class="icon mr-2"><i class="fas fa-database"></i></span>
Datasets
</p>
</div>
<div class="card-content">
<div class="content">
<p>We carefully selected datasets based on:</p>
<ul style="list-style-type: none;">
<li><strong>Domain relevance:</strong> Majority of content directly related to finance</li>
<li><strong>Licensing:</strong> Fair usage licensing and proper attribution</li>
<li><strong>Quality:</strong> Transparent sourcing with minimal risk of label corruption</li>
<li><strong>Complexity:</strong> Exercises real financial knowledge, not trivial tasks</li>
</ul>
</div>
</div>
</div>
</div>
<!-- Models -->
<div class="column is-6">
<div class="card h-100">
<div class="card-header">
<p class="card-header-title">
<span class="icon mr-2"><i class="fas fa-robot"></i></span>
Models
</p>
</div>
<div class="card-content">
<div class="content">
<p>We evaluated diverse language models across two categories:</p>
<div class="columns">
<div class="column is-half">
<h6 class="title is-6">Proprietary</h6>
<ul style="list-style-type: none;">
<li>GPT-4o & o1-mini</li>
<li>Gemini-1.5 Pro</li>
<li>Claude 3.5 Sonnet</li>
<li>Claude 3 Haiku</li>
<li>Cohere Command (R 7B, R+)</li>
<li>Jamba 1.5 Mini & Large</li>
</ul>
</div>
<div class="column is-half">
<h6 class="title is-6">Open-weight</h6>
<ul style="list-style-type: none;">
<li>Llama-3 (8B, 70B)</li>
<li>DeepSeek V3 & R-1</li>
<li>DBRX Instruct</li>
<li>Qwen-2 (72B) & QwQ-32B</li>
<li>Mistral (7B, 8x7B, 8x22B)</li>
<li>Gemma-2 (9B, 27B)</li>
<li>WizardLM-2 (8x22B)</li>
</ul>
</div>
</div>
</div>
</div>
</div>
</div>
<!-- Evaluation Process -->
<div class="column is-6">
<div class="card h-100">
<div class="card-header">
<p class="card-header-title">
<span class="icon mr-2"><i class="fas fa-chart-bar"></i></span>
Evaluation Process
</p>
</div>
<div class="card-content">
<div class="content">
<p>Our two-stage evaluation approach includes:</p>
<ol>
<li><strong>Generation:</strong> Language model generates responses to task-specific inputs</li>
<li><strong>Extraction:</strong> Separate process identifies relevant output using structured pattern matching</li>
</ol>
<p>Pipeline stages:</p>
<ul style="list-style-type: none;">
<li>Configuration of tasks, datasets, and parameters</li>
<li>Model interaction via local instantiation or API</li>
<li>Post-processing and structured output extraction</li>
<li>Task-specific metric computation and logging</li>
</ul>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
</div>
<!--/ Methodology -->
</div>
</section>
<section id="results" class="section content-section">
<div class="container is-max-desktop">
<!-- Results -->
<div class="columns is-centered">
<div class="column is-full">
<h2 class="title is-3 section-title has-text-centered">Results</h2>
<div class="content">
<div class="notification is-info is-light mb-5">
<p class="is-size-5 has-text-centered mb-0">
Our comprehensive evaluation reveals significant performance variations across different financial tasks and models.
</p>
</div>
</div>
<section class="section">
<div class="container">
<h1 class="title is-2">FLaME: Financial Language Model Evaluation Results</h1>
<div class="content">
<p class="is-size-5">
This page presents the results of the FLaME evaluation across various financial NLP tasks.
Each tab shows performance metrics for different task categories.
</p>
</div>
<div class="tabs-container">
<div class="tabs is-centered is-boxed">
<ul>
<li class="is-active" data-tab="main">
<a>
<span>All Tasks</span>
</a>
</li>
<li data-tab="causal-analysis">
<a>
<span>Causal Analysis</span>
</a>
</li>
<li data-tab="information-retrieval">
<a>
<span>Information Retrieval</span>
</a>
</li>
<li data-tab="question-answering">
<a>
<span>Question Answering</span>
</a>
</li>
<li data-tab="sentiment-analysis">
<a>
<span>Sentiment Analysis</span>
</a>
</li>
<li data-tab="text-classification">
<a>
<span>Text Classification</span>
</a>
</li>
<li data-tab="text-summarization">
<a>
<span>Text Summarization</span>
</a>
</li>
<li data-tab="cost-analysis">
<a>
<span>Cost Analysis</span>
</a>
</li>
</ul>
</div>
<div id="main" class="tab-content">
<h2 class="title is-4">Overall Performance Across All Tasks</h2>
<div class="results-table">
<table class="table is-bordered is-striped is-narrow is-hoverable is-fullwidth">
<thead>
<tr>
<th>Model</th>
<th colspan="4" class="has-text-centered column-border-left">Information Retrieval</th>
<th class="has-text-centered column-border-left">*</th>
<th colspan="3" class="has-text-centered column-border-left">Sentiment Analysis</th>
<th colspan="2" class="has-text-centered column-border-left">Causal Analysis</th>
<th colspan="5" class="has-text-centered column-border-left">Text Classification</th>
<th colspan="3" class="has-text-centered column-border-left">Question Answering</th>
<th colspan="2" class="has-text-centered column-border-left">Summarization</th>
</tr>
<tr>
<th>Dataset</th>
<th class="has-text-centered tooltip-trigger column-border-left" data-title="FiNER-ORD" data-tooltip="FiNER-ORD is a manually annotated named entity recognition dataset comprising financial news articles with detailed entity annotations. The task requires identifying and correctly classifying person, location, and organization entities in financial contexts.">FiNER<span class="metric-label">F1</span></th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="FinRED" data-tooltip="FinRED is a specialized relation extraction dataset created from financial news and earnings call transcripts using distance supervision based on Wikidata triplets. The task involves identifying and extracting financial relationships between entities to understand connections in financial contexts.">FR<span class="metric-label">F1</span></th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="ReFiND" data-tooltip="ReFiND is a comprehensive relation extraction dataset containing approximately 29,000 annotated instances with 22 distinct relation types across 8 entity pair categories from various financial documents. The task requires identifying specific relationships between financial entities in complex documents like SEC filings.">RD<span class="metric-label">F1</span></th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="FNXL" data-tooltip="FNXL contains 79,088 sentences with 142,922 annotated numerals extracted from SEC 10-K reports and categorized under 2,794 distinct numerical labels. The information extraction task requires identifying, categorizing and understanding the financial significance of numerical entities in regulatory filings.">FNXL<span class="metric-label">F1</span></th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="FinEntity" data-tooltip="FinEntity consists of 979 financial news paragraphs containing 2,131 manually-annotated financial entities with sentiment classifications. The task involves identifying companies and asset classes in financial texts while determining the associated sentiment expressed toward each entity.">FE<span class="metric-label">F1</span></th>
<th class="has-text-centered tooltip-trigger column-border-left" data-title="FiQA Task 1" data-tooltip="FiQA Task 1 focuses on aspect-based financial sentiment analysis in microblog posts and news headlines using a continuous scale from -1 (negative) to 1 (positive). The regression task requires models to accurately predict the sentiment score that reflects investor perception of financial texts.">FiQA<span class="metric-label">MSE</span></th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="SubjECTive-QA" data-tooltip="SubjECTive-QA contains 49,446 annotations across 2,747 question-answer pairs extracted from 120 earnings call transcripts. The multi-label classification task involves analyzing six subjective features in financial discourse: assertiveness, cautiousness, optimism, specificity, clarity, and relevance.">SQA<span class="metric-label">F1</span></th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="Financial Phrase Bank" data-tooltip="Financial Phrase Bank (FPB) contains 4,840 sentences from financial news articles categorized as positive, negative, or neutral by 16 finance experts using majority voting. The sentiment classification task requires understanding how these statements might influence investor perception of stock prices.">FPB<span class="metric-label">F1</span></th>
<th class="has-text-centered tooltip-trigger column-border-left" data-title="FinCausal (CD)" data-tooltip="FinCausal Causal Discovery (CD) contains 29,444 text sections from financial news, with 2,136 annotated as expressing causal relationships. The task involves extracting precise cause and effect spans from financial texts that contain causal relationships.">CD<span class="metric-label">F1</span></th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="FinCausal (CC)" data-tooltip="FinCausal Causality Classification (CC) consists of 29,444 text sections from financial news with binary annotations indicating causal relationships. The classification task requires determining whether a given financial text section contains a causal relationship (1) or not (0).">CC<span class="metric-label">F1</span></th>
<th class="has-text-centered tooltip-trigger column-border-left" data-title="Banking77" data-tooltip="Banking77 is a fine-grained dataset comprising 13,083 customer service queries annotated with 77 unique intents from the banking domain. The task involves accurately classifying each customer query into the correct intent category to improve automated banking support systems.">B77<span class="metric-label">F1</span></th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="FinBench" data-tooltip="FinBench is a comprehensive evaluation dataset containing 333,000 labeled instances that combines tabular data and profile text for financial risk prediction. The task requires models to predict financial outcomes across three key risk categories: default, fraud, and customer churn.">FB<span class="metric-label">F1</span></th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="FOMC" data-tooltip="FOMC is a dataset containing Federal Open Market Committee speeches, meeting minutes, and press conference transcripts spanning from 1996 to 2022. The classification task involves determining whether the monetary policy stance expressed in each document is hawkish (tighter monetary policy) or dovish (looser monetary policy).">FOMC<span class="metric-label">F1</span></th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="NumClaim" data-tooltip="NumClaim is an expert-annotated dataset for detecting and analyzing fine-grained investor claims within financial narratives that contain numerical information. The task requires identifying and categorizing claims containing numerals in analyst reports and earnings call transcripts for investment decision making.">NC<span class="metric-label">F1</span></th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="Headlines" data-tooltip="Headlines is a dataset containing 11,412 human-annotated financial news headlines focused on commodities, particularly gold, spanning from 2000 to 2019. The classification task involves identifying binary indicators for price mentions and directional price movements in these concise financial texts.">HL<span class="metric-label">Acc</span></th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="ConvFinQA" data-tooltip="ConvFinQA is a multi-turn question answering dataset with 3,892 conversations containing 14,115 questions that explore chains of numerical reasoning in financial contexts. The conversational task requires maintaining context while performing sequential numerical operations to answer increasingly complex financial questions.">CFQA<span class="metric-label">Acc</span></th>
<th class="has-text-centered tooltip-trigger column-border-left" data-title="FinQA" data-tooltip="FinQA contains 8,281 question-answer pairs derived from financial reports that require numerical reasoning over tabular financial data. The question-answering task features multi-step reasoning challenges with full annotation of reasoning programs to solve complex financial queries.">FinQA<span class="metric-label">Acc</span></th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="TATQA" data-tooltip="TATQA is a large-scale question answering dataset for hybrid data sources that combines tables and text from financial reports. The task emphasizes numerical reasoning operations across multiple formats, requiring models to integrate information from structured and unstructured sources to answer financial questions.">TQA<span class="metric-label">Acc</span></th>
<th class="has-text-centered tooltip-trigger tooltip-right column-border-left" data-title="ECTSum" data-tooltip="ECTSum contains 2,425 document-summary pairs featuring earnings call transcripts paired with concise bullet-point summaries extracted from Reuters articles. The summarization task requires extracting and condensing key financial information from lengthy corporate communications into brief, informative points.">ECTSum<span class="metric-label">BERT-F1</span></th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="EDTSum" data-tooltip="EDTSum consists of 2,000 financial news articles paired with their headlines as ground-truth summaries for evaluating text summarization. The task challenges models to condense complex financial news articles into concise, informative headlines that capture the essential information.">EDTSum<span class="metric-label">BERT-F1</span></th>
</tr>
</thead>
<tbody>
<tr>
<td>Llama 3 70B Instruct</td>
<td class="column-border-left">.701</td><td>.332</td><td>.883</td><td>.020</td>
<td class="column-border-left">.469</td><td>.123</td><td>.535</td><td>.902</td>
<td class="column-border-left">.142</td><td>.192</td>
<td class="column-border-left">.645</td><td>.309</td><td>.652</td><td>.386</td><td>.811</td>
<td class="column-border-left">.709</td><td>.809</td><td>.772</td>
<td class="column-border-left">.754</td><td class="performance-medium">.817</td>
</tr>
<tr>
<td>Llama 3 8B Instruct</td>
<td class="column-border-left">.565</td><td>.289</td><td>.705</td><td>.003</td>
<td class="column-border-left">.350</td><td>.161</td><td class="performance-best">.600</td><td>.698</td>
<td class="column-border-left">.049</td><td>.234</td>
<td class="column-border-left">.512</td><td>.659</td><td>.497</td><td>.511</td><td>.763</td>
<td class="column-border-left">.268</td><td>.767</td><td>.706</td>
<td class="column-border-left">.757</td><td>.811</td>
</tr>
<tr>
<td>DBRX Instruct</td>
<td class="column-border-left">.489</td><td>.304</td><td>.778</td><td>.009</td>
<td class="column-border-left">.006</td><td>.160</td><td>.436</td><td>.499</td>
<td class="column-border-left">.087</td><td>.231</td>
<td class="column-border-left">.574</td><td>.483</td><td>.193</td><td>.319</td><td>.746</td>
<td class="column-border-left">.252</td><td>.738</td><td>.633</td>
<td class="column-border-left">.729</td><td>.806</td>
</tr>
<tr>
<td>DeepSeek LLM (67B)</td>
<td class="column-border-left">.745</td><td>.334</td><td>.879</td><td>.007</td>
<td class="column-border-left">.416</td><td>.118</td><td>.462</td><td>.811</td>
<td class="column-border-left">.025</td><td>.193</td>
<td class="column-border-left">.578</td><td>.492</td><td>.407</td><td>.151</td><td>.778</td>
<td class="column-border-left">.174</td><td>.742</td><td>.355</td>
<td class="column-border-left">.681</td><td>.807</td>
</tr>
<tr>
<td>Gemma 2 27B</td>
<td class="column-border-left">.761</td><td>.356</td><td>.902</td><td>.006</td>
<td class="column-border-left">.298</td><td class="performance-best">.100</td><td>.515</td><td>.884</td>
<td class="column-border-left">.133</td><td>.242</td>
<td class="column-border-left">.621</td><td>.538</td><td>.620</td><td>.408</td><td>.808</td>
<td class="column-border-left">.268</td><td>.768</td><td>.734</td>
<td class="column-border-left">.723</td><td>.814</td>
</tr>
<tr>
<td>Gemma 2 9B</td>
<td class="column-border-left">.651</td><td>.331</td><td>.892</td><td>.005</td>
<td class="column-border-left">.367</td><td>.189</td><td>.491</td><td class="performance-medium">.940</td>
<td class="column-border-left">.105</td><td>.207</td>
<td class="column-border-left">.609</td><td>.541</td><td>.519</td><td>.365</td><td class="performance-best">.856</td>
<td class="column-border-left">.292</td><td>.779</td><td>.750</td>
<td class="column-border-left">.585</td><td class="performance-medium">.817</td>
</tr>
<tr>
<td>Mistral (7B) Instruct v0.3</td>
<td class="column-border-left">.526</td><td>.276</td><td>.771</td><td>.004</td>
<td class="column-border-left">.368</td><td>.135</td><td>.522</td><td>.841</td>
<td class="column-border-left">.052</td><td>.227</td>
<td class="column-border-left">.528</td><td>.503</td><td>.542</td><td>.412</td><td>.779</td>
<td class="column-border-left">.199</td><td>.655</td><td>.553</td>
<td class="column-border-left">.750</td><td>.811</td>
</tr>
<tr>
<td>Mixtral-8x22B Instruct</td>
<td class="column-border-left">.635</td><td>.367</td><td>.811</td><td>.009</td>
<td class="column-border-left">.435</td><td>.221</td><td>.510</td><td>.776</td>
<td class="column-border-left">.125</td><td class="performance-best">.308</td>
<td class="column-border-left">.602</td><td>.221</td><td>.465</td><td>.513</td><td class="performance-low">.835</td>
<td class="column-border-left">.285</td><td>.766</td><td>.666</td>
<td class="column-border-left">.758</td><td>.815</td>
</tr>
<tr>
<td>Mixtral-8x7B Instruct</td>
<td class="column-border-left">.598</td><td>.282</td><td>.845</td><td>.009</td>
<td class="column-border-left">.267</td><td>.208</td><td>.498</td><td>.893</td>
<td class="column-border-left">.055</td><td>.229</td>
<td class="column-border-left">.547</td><td>.396</td><td>.603</td><td>.583</td><td>.805</td>
<td class="column-border-left">.315</td><td>.611</td><td>.501</td>
<td class="column-border-left">.747</td><td>.810</td>
</tr>
<tr>
<td>Qwen 2 Instruct (72B)</td>
<td class="column-border-left">.748</td><td>.348</td><td>.854</td><td>.012</td>
<td class="column-border-left">.483</td><td>.205</td><td>.576</td><td>.901</td>
<td class="column-border-left">.190</td><td>.184</td>
<td class="column-border-left">.627</td><td>.495</td><td>.605</td><td>.639</td><td>.830</td>
<td class="column-border-left">.269</td><td>.819</td><td>.715</td>
<td class="column-border-left">.752</td><td>.811</td>
</tr>
<tr>
<td>WizardLM-2 8x22B</td>
<td class="column-border-left">.744</td><td>.355</td><td>.852</td><td>.008</td>
<td class="column-border-left">.226</td><td>.129</td><td>.566</td><td>.779</td>
<td class="column-border-left">.114</td><td>.201</td>
<td class="column-border-left">.648</td><td>.500</td><td>.505</td><td>.272</td><td>.797</td>
<td class="column-border-left">.247</td><td>.796</td><td>.725</td>
<td class="column-border-left">.735</td><td>.808</td>
</tr>
<tr>
<td>DeepSeek-V3</td>
<td class="column-border-left performance-low">.790</td><td class="performance-medium">.437</td><td>.934</td><td class="performance-low">.045</td>
<td class="column-border-left">.549</td><td>.150</td><td class="performance-low">.583</td><td>.814</td>
<td class="column-border-left performance-low">.198</td><td>.170</td>
<td class="column-border-left performance-medium">.714</td><td>.487</td><td>.578</td><td>.675</td><td>.729</td>
<td class="column-border-left">.261</td><td class="performance-medium">.840</td><td class="performance-low">.779</td>
<td class="column-border-left">.750</td><td>.815</td>
</tr>
<tr>
<td><strong>DeepSeek R1</strong></td>
<td class="column-border-left performance-best">.807</td><td>.393</td><td class="performance-best">.952</td><td class="performance-best">.057</td>
<td class="column-border-left performance-low">.587</td><td>.110</td><td>.499</td><td>.902</td>
<td class="column-border-left performance-best">.337</td><td>.202</td>
<td class="column-border-left performance-best">.763</td><td>.419</td><td class="performance-medium">.670</td><td>.688</td><td>.769</td>
<td class="column-border-left performance-best">.853</td><td class="performance-low">.836</td><td class="performance-best">.858</td>
<td class="column-border-left">.759</td><td>.804</td>
</tr>
<tr>
<td><strong>QwQ-32B-Preview</strong></td>
<td class="column-border-left">.685</td><td>.270</td><td>.656</td><td>.001</td>
<td class="column-border-left">.005</td><td>.141</td><td>.550</td><td>.815</td>
<td class="column-border-left">.131</td><td>.220</td>
<td class="column-border-left">.613</td><td class="performance-medium">.784</td><td>.555</td><td>.020</td><td>.744</td>
<td class="column-border-left">.282</td><td>.793</td><td class="performance-medium">.796</td>
<td class="column-border-left">.696</td><td class="performance-medium">.817</td>
</tr>
<tr>
<td>Jamba 1.5 Mini</td>
<td class="column-border-left">.552</td><td>.284</td><td>.844</td><td>.005</td>
<td class="column-border-left">.132</td><td>.119</td><td>.418</td><td>.765</td>
<td class="column-border-left">.043</td><td class="performance-medium">.270</td>
<td class="column-border-left">.508</td><td class="performance-best">.898</td><td>.499</td><td>.151</td><td>.682</td>
<td class="column-border-left">.218</td><td>.666</td><td>.586</td>
<td class="column-border-left">.741</td><td>.816</td>
</tr>
<tr>
<td>Jamba 1.5 Large</td>
<td class="column-border-left">.693</td><td>.341</td><td>.862</td><td>.005</td>
<td class="column-border-left">.397</td><td>.183</td><td>.582</td><td>.798</td>
<td class="column-border-left">.074</td><td>.176</td>
<td class="column-border-left">.628</td><td>.618</td><td>.550</td><td>.541</td><td>.782</td>
<td class="column-border-left">.225</td><td>.790</td><td>.660</td>
<td class="column-border-left">.734</td><td class="performance-best">.818</td>
</tr>
<tr>
<td>Claude 3.5 Sonnet</td>
<td class="column-border-left performance-medium">.799</td><td class="performance-best">.439</td><td>.891</td><td class="performance-medium">.047</td>
<td class="column-border-left performance-medium">.655</td><td class="performance-medium">.101</td><td>.553</td><td class="performance-best">.944</td>
<td class="column-border-left">.196</td><td>.197</td>
<td class="column-border-left">.668</td><td>.634</td><td class="performance-best">.674</td><td class="performance-low">.692</td><td>.827</td>
<td class="column-border-left">.402</td><td class="performance-best">.844</td><td>.700</td>
<td class="column-border-left performance-low">.767</td><td>.813</td>
</tr>
<tr>
<td>Claude 3 Haiku</td>
<td class="column-border-left">.711</td><td>.285</td><td>.883</td><td>.015</td>
<td class="column-border-left">.494</td><td>.167</td><td>.463</td><td>.908</td>
<td class="column-border-left">.081</td><td>.200</td>
<td class="column-border-left">.622</td><td>.022</td><td>.631</td><td>.558</td><td>.781</td>
<td class="column-border-left">.421</td><td>.803</td><td>.733</td>
<td class="column-border-left">.646</td><td>.808</td>
</tr>
<tr>
<td>Cohere Command R 7B</td>
<td class="column-border-left">.748</td><td>.194</td><td>.845</td><td>.018</td>
<td class="column-border-left">.441</td><td>.164</td><td>.532</td><td>.840</td>
<td class="column-border-left">.057</td><td class="performance-low">.255</td>
<td class="column-border-left">.516</td><td class="performance-low">.762</td><td>.459</td><td>.068</td><td>.770</td>
<td class="column-border-left">.212</td><td>.709</td><td>.716</td>
<td class="column-border-left">.750</td><td>.815</td>
</tr>
<tr>
<td>Cohere Command R +</td>
<td class="column-border-left">.756</td><td>.333</td><td>.922</td><td>.021</td>
<td class="column-border-left">.452</td><td class="performance-low">.106</td><td>.533</td><td>.699</td>
<td class="column-border-left">.080</td><td>.238</td>
<td class="column-border-left">.651</td><td>.684</td><td>.393</td><td>.118</td><td>.812</td>
<td class="column-border-left">.259</td><td>.776</td><td>.698</td>
<td class="column-border-left">.751</td><td>.810</td>
</tr>
<tr>
<td>Google Gemini 1.5 Pro</td>
<td class="column-border-left">.712</td><td>.374</td><td class="performance-medium">.944</td><td>.019</td>
<td class="column-border-left">.393</td><td>.144</td><td class="performance-medium">.593</td><td>.885</td>
<td class="column-border-left">.196</td><td>.217</td>
<td class="column-border-left">.418</td><td>.336</td><td>.579</td><td>.525</td><td class="performance-medium">.837</td>
<td class="column-border-left">.280</td><td>.829</td><td>.763</td>
<td class="column-border-left performance-best">.777</td><td class="performance-medium">.817</td>
</tr>
<tr>
<td>OpenAI gpt-4o</td>
<td class="column-border-left">.766</td><td>.399</td><td class="performance-low">.942</td><td>.037</td>
<td class="column-border-left">.523</td><td>.184</td><td>.541</td><td class="performance-low">.928</td>
<td class="column-border-left">.130</td><td>.222</td>
<td class="column-border-left performance-low">.710</td><td>.524</td><td class="performance-low">.664</td><td class="performance-best">.750</td><td>.824</td>
<td class="column-border-left performance-low">.749</td><td>.836</td><td>.754</td>
<td class="column-border-left performance-medium">.773</td><td>.816</td>
</tr>
<tr>
<td><strong>OpenAI o1-mini</strong></td>
<td class="column-border-left">.761</td><td class="performance-low">.403</td><td>.876</td><td>.010</td>
<td class="column-border-left performance-best">.662</td><td>.120</td><td>.542</td><td>.917</td>
<td class="column-border-left performance-medium">.289</td><td>.209</td>
<td class="column-border-left">.670</td><td>.612</td><td>.635</td><td class="performance-medium">.720</td><td>.769</td>
<td class="column-border-left performance-medium">.840</td><td>.799</td><td>.698</td>
<td class="column-border-left">.763</td><td>.816</td>
</tr>
</tbody>
</table>
<div class="content is-small mt-4">
<p><strong>Note:</strong> Color highlighting indicates performance ranking:
<span class="performance-best"> Best </span>,
<span class="performance-medium"> Strong </span>,
<span class="performance-low"> Good </span>
</p>
</div>
</div>
</div>
<!-- Causal Analysis tab content -->
<div id="causal-analysis" class="tab-content">
<h2 class="title is-4">Causal Analysis Task Results</h2>
<div class="results-table">
<table class="table is-bordered is-striped is-narrow is-hoverable is-fullwidth">
<thead>
<tr>
<th rowspan="2">Model</th>
<th colspan="4" class="has-text-centered tooltip-trigger tooltip-right" data-title="FinCausal (CD)" data-tooltip="FinCausal Causal Discovery (CD) contains 29,444 text sections from financial news, with 2,136 annotated as expressing causal relationships. The task involves extracting precise cause and effect spans from financial texts that contain causal relationships.">Causal Detection (CD)</th>
<th colspan="4" class="has-text-centered tooltip-trigger tooltip-right" data-title="FinCausal (CC)" data-tooltip="FinCausal Causality Classification (CC) consists of 29,444 text sections from financial news with binary annotations indicating causal relationships. The classification task requires determining whether a given financial text section contains a causal relationship (1) or not (0).">Causal Classification (CC)</th>
</tr>
<tr>
<th class="has-text-centered">Accuracy</th>
<th class="has-text-centered">Precision</th>
<th class="has-text-centered">Recall</th>
<th class="has-text-centered">F1</th>
<th class="has-text-centered">Precision</th>
<th class="has-text-centered">Recall</th>
<th class="has-text-centered">F1</th>
<th class="has-text-centered">Accuracy</th>
</tr>
</thead>
<tbody>
<tr>
<td>Llama 3 70B Instruct</td>
<td class="has-text-centered">0.148</td>
<td class="has-text-centered">0.429</td>
<td class="has-text-centered">0.148</td>
<td class="has-text-centered">0.142</td>
<td class="has-text-centered">0.241</td>
<td class="has-text-centered">0.329</td>
<td class="has-text-centered">0.192</td>
<td class="has-text-centered">0.198</td>
</tr>
<tr>
<td>Llama 3 8B Instruct</td>
<td class="has-text-centered">0.097</td>
<td class="has-text-centered">0.341</td>
<td class="has-text-centered">0.097</td>
<td class="has-text-centered">0.049</td>
<td class="has-text-centered">0.232</td>
<td class="has-text-centered">0.241</td>
<td class="has-text-centered">0.234</td>
<td class="has-text-centered performance-medium">0.380</td>
</tr>
<tr>
<td>DBRX Instruct</td>
<td class="has-text-centered">0.078</td>
<td class="has-text-centered">0.521</td>
<td class="has-text-centered">0.078</td>
<td class="has-text-centered">0.087</td>
<td class="has-text-centered">0.276</td>
<td class="has-text-centered">0.313</td>
<td class="has-text-centered">0.231</td>
<td class="has-text-centered">0.235</td>
</tr>
<tr>
<td>DeepSeek LLM (67B)</td>
<td class="has-text-centered">0.026</td>
<td class="has-text-centered">0.214</td>
<td class="has-text-centered">0.026</td>
<td class="has-text-centered">0.025</td>
<td class="has-text-centered">0.141</td>
<td class="has-text-centered">0.328</td>
<td class="has-text-centered">0.193</td>
<td class="has-text-centered">0.221</td>
</tr>
<tr>
<td>Gemma 2 27B</td>
<td class="has-text-centered">0.115</td>
<td class="has-text-centered">0.510</td>
<td class="has-text-centered">0.115</td>
<td class="has-text-centered">0.133</td>
<td class="has-text-centered">0.309</td>
<td class="has-text-centered">0.310</td>
<td class="has-text-centered">0.242</td>
<td class="has-text-centered">0.262</td>
</tr>
<tr>
<td>Gemma 2 9B</td>
<td class="has-text-centered">0.115</td>
<td class="has-text-centered">0.394</td>
<td class="has-text-centered">0.115</td>
<td class="has-text-centered">0.105</td>
<td class="has-text-centered">0.275</td>
<td class="has-text-centered">0.294</td>
<td class="has-text-centered">0.207</td>
<td class="has-text-centered">0.258</td>
</tr>
<tr>
<td>Mistral (7B) Instruct v0.3</td>
<td class="has-text-centered">0.078</td>
<td class="has-text-centered">0.455</td>
<td class="has-text-centered">0.078</td>
<td class="has-text-centered">0.052</td>
<td class="has-text-centered">0.339</td>
<td class="has-text-centered performance-best">0.361</td>
<td class="has-text-centered">0.227</td>
<td class="has-text-centered">0.258</td>
</tr>
<tr>
<td>Mixtral-8x22B Instruct</td>
<td class="has-text-centered">0.131</td>
<td class="has-text-centered">0.486</td>
<td class="has-text-centered">0.131</td>
<td class="has-text-centered">0.125</td>
<td class="has-text-centered">0.344</td>
<td class="has-text-centered">0.310</td>
<td class="has-text-centered performance-best">0.308</td>
<td class="has-text-centered performance-medium">0.318</td>
</tr>
<tr>
<td>Mixtral-8x7B Instruct</td>
<td class="has-text-centered">0.088</td>
<td class="has-text-centered">0.510</td>
<td class="has-text-centered">0.088</td>
<td class="has-text-centered">0.055</td>
<td class="has-text-centered">0.308</td>
<td class="has-text-centered">0.314</td>
<td class="has-text-centered">0.229</td>
<td class="has-text-centered">0.273</td>
</tr>
<tr>
<td>Qwen 2 Instruct (72B)</td>
<td class="has-text-centered">0.139</td>
<td class="has-text-centered">0.489</td>
<td class="has-text-centered">0.139</td>
<td class="has-text-centered">0.190</td>
<td class="has-text-centered">0.208</td>
<td class="has-text-centered">0.330</td>
<td class="has-text-centered">0.184</td>
<td class="has-text-centered">0.188</td>
</tr>
<tr>
<td>WizardLM-2 8x22B</td>
<td class="has-text-centered">0.076</td>
<td class="has-text-centered">0.453</td>
<td class="has-text-centered">0.076</td>
<td class="has-text-centered">0.114</td>
<td class="has-text-centered">0.263</td>
<td class="has-text-centered">0.347</td>
<td class="has-text-centered">0.201</td>
<td class="has-text-centered">0.237</td>
</tr>
<tr>
<td>DeepSeek-V3</td>
<td class="has-text-centered">0.164</td>
<td class="has-text-centered">0.528</td>
<td class="has-text-centered">0.164</td>
<td class="has-text-centered performance-medium">0.198</td>
<td class="has-text-centered">0.194</td>
<td class="has-text-centered">0.327</td>
<td class="has-text-centered">0.170</td>
<td class="has-text-centered">0.248</td>
</tr>
<tr>
<td>DeepSeek R1</td>
<td class="has-text-centered performance-best">0.245</td>
<td class="has-text-centered performance-medium">0.643</td>
<td class="has-text-centered performance-best">0.245</td>
<td class="has-text-centered performance-best">0.337</td>
<td class="has-text-centered performance-best">0.385</td>
<td class="has-text-centered">0.318</td>
<td class="has-text-centered">0.202</td>
<td class="has-text-centered">0.221</td>
</tr>
<tr>
<td>QwQ-32B-Preview</td>
<td class="has-text-centered">0.110</td>
<td class="has-text-centered">0.473</td>
<td class="has-text-centered">0.110</td>
<td class="has-text-centered">0.131</td>
<td class="has-text-centered">0.193</td>
<td class="has-text-centered">0.262</td>
<td class="has-text-centered">0.220</td>
<td class="has-text-centered performance-best">0.465</td>
</tr>
<tr>
<td>Jamba 1.5 Mini</td>
<td class="has-text-centered">0.050</td>
<td class="has-text-centered">0.280</td>
<td class="has-text-centered">0.050</td>
<td class="has-text-centered">0.043</td>
<td class="has-text-centered">0.323</td>
<td class="has-text-centered">0.283</td>
<td class="has-text-centered performance-medium">0.270</td>
<td class="has-text-centered">0.295</td>
</tr>
<tr>
<td>Jamba 1.5 Large</td>
<td class="has-text-centered">0.076</td>
<td class="has-text-centered">0.517</td>
<td class="has-text-centered">0.076</td>
<td class="has-text-centered">0.074</td>
<td class="has-text-centered">0.268</td>
<td class="has-text-centered">0.248</td>
<td class="has-text-centered">0.176</td>
<td class="has-text-centered">0.200</td>
</tr>
<tr>
<td>Claude 3.5 Sonnet</td>
<td class="has-text-centered">0.154</td>
<td class="has-text-centered">0.564</td>
<td class="has-text-centered">0.154</td>
<td class="has-text-centered">0.196</td>
<td class="has-text-centered">0.259</td>
<td class="has-text-centered">0.336</td>
<td class="has-text-centered">0.197</td>
<td class="has-text-centered">0.235</td>
</tr>
<tr>
<td>Claude 3 Haiku</td>
<td class="has-text-centered">0.082</td>
<td class="has-text-centered">0.388</td>
<td class="has-text-centered">0.082</td>
<td class="has-text-centered">0.081</td>
<td class="has-text-centered performance-medium">0.369</td>
<td class="has-text-centered">0.347</td>
<td class="has-text-centered">0.200</td>
<td class="has-text-centered">0.203</td>
</tr>
<tr>
<td>Cohere Command R 7B</td>
<td class="has-text-centered">0.089</td>
<td class="has-text-centered">0.363</td>
<td class="has-text-centered">0.089</td>
<td class="has-text-centered">0.057</td>
<td class="has-text-centered performance-medium">0.379</td>
<td class="has-text-centered performance-medium">0.356</td>
<td class="has-text-centered performance-medium">0.255</td>
<td class="has-text-centered">0.275</td>
</tr>
<tr>
<td>Cohere Command R +</td>
<td class="has-text-centered">0.090</td>
<td class="has-text-centered">0.453</td>
<td class="has-text-centered">0.090</td>
<td class="has-text-centered">0.080</td>
<td class="has-text-centered">0.353</td>
<td class="has-text-centered">0.336</td>
<td class="has-text-centered">0.238</td>
<td class="has-text-centered">0.265</td>
</tr>
<tr>
<td>Google Gemini 1.5 Pro</td>
<td class="has-text-centered performance-medium">0.165</td>
<td class="has-text-centered">0.514</td>
<td class="has-text-centered performance-medium">0.165</td>
<td class="has-text-centered">0.196</td>
<td class="has-text-centered">0.265</td>
<td class="has-text-centered performance-medium">0.357</td>
<td class="has-text-centered">0.217</td>
<td class="has-text-centered">0.258</td>
</tr>
<tr>
<td>OpenAI gpt-4o</td>
<td class="has-text-centered">0.082</td>
<td class="has-text-centered performance-medium">0.576</td>
<td class="has-text-centered">0.082</td>
<td class="has-text-centered">0.130</td>
<td class="has-text-centered">0.254</td>
<td class="has-text-centered">0.327</td>
<td class="has-text-centered">0.222</td>
<td class="has-text-centered">0.235</td>
</tr>
<tr>
<td>OpenAI o1-mini</td>
<td class="has-text-centered performance-medium">0.206</td>
<td class="has-text-centered performance-best">0.648</td>
<td class="has-text-centered performance-medium">0.206</td>
<td class="has-text-centered performance-medium">0.289</td>
<td class="has-text-centered">0.325</td>
<td class="has-text-centered">0.316</td>
<td class="has-text-centered">0.209</td>
<td class="has-text-centered">0.233</td>
</tr>
</tbody>
</table>
<div class="content is-small mt-4">
<p><strong>Note:</strong> Color highlighting indicates performance ranking:
<span class="performance-best"> Best </span>,
<span class="performance-medium"> Strong </span>,
<span class="performance-low"> Good </span>
</p>
</div>
</div>
</div>
<!-- Information Retrieval tab content -->
<div id="information-retrieval" class="tab-content">
<h2 class="title is-4">Information Retrieval Task Results</h2>
<div class="results-table">
<table class="table is-bordered is-striped is-narrow is-hoverable is-fullwidth">
<thead>
<tr>
<th rowspan="2">Model</th>
<th colspan="4" class="has-text-centered tooltip-trigger tooltip-right" data-title="FiNER-ORD" data-tooltip="FiNER-ORD is a manually annotated named entity recognition dataset comprising financial news articles with detailed entity annotations. The task requires identifying and correctly classifying person, location, and organization entities in financial contexts.">FiNER-ORD</th>
<th colspan="4" class="has-text-centered tooltip-trigger tooltip-right" data-title="FinRED" data-tooltip="FinRED is a specialized relation extraction dataset created from financial news and earnings call transcripts using distance supervision based on Wikidata triplets. The task involves identifying and extracting financial relationships between entities to understand connections in financial contexts.">FinRED</th>
<th colspan="4" class="has-text-centered tooltip-trigger tooltip-right" data-title="ReFiND" data-tooltip="ReFiND is a comprehensive relation extraction dataset containing approximately 29,000 annotated instances with 22 distinct relation types across 8 entity pair categories from various financial documents. The task requires identifying specific relationships between financial entities in complex documents like SEC filings.">ReFiND</th>
<th colspan="4" class="has-text-centered tooltip-trigger tooltip-right" data-title="FNXL" data-tooltip="FNXL is a financial news cross-lingual dataset containing 1,000 news articles in 5 languages with 5,000 annotated entity mentions. The task requires identifying and classifying entities in financial news articles across multiple languages.">FNXL</th>
<th colspan="4" class="has-text-centered tooltip-trigger tooltip-right" data-title="FinEntity" data-tooltip="FinEntity consists of 979 financial news paragraphs containing 2,131 manually-annotated financial entities with sentiment classifications. The task involves identifying companies and asset classes in financial texts while determining the associated sentiment expressed toward each entity.">FinEntity</th>
</tr>
<tr>
<th class="has-text-centered">Precision</th>
<th class="has-text-centered">Recall</th>
<th class="has-text-centered">F1</th>
<th class="has-text-centered">Accuracy</th>
<th class="has-text-centered">Accuracy</th>
<th class="has-text-centered">Precision</th>
<th class="has-text-centered">Recall</th>
<th class="has-text-centered">F1</th>
<th class="has-text-centered">Accuracy</th>
<th class="has-text-centered">Precision</th>
<th class="has-text-centered">Recall</th>
<th class="has-text-centered">F1</th>
<th class="has-text-centered">Precision</th>
<th class="has-text-centered">Recall</th>
<th class="has-text-centered">F1</th>
<th class="has-text-centered">Accuracy</th>
<th class="has-text-centered">Precision</th>
<th class="has-text-centered">Recall</th>
<th class="has-text-centered">Accuracy</th>
<th class="has-text-centered">F1</th>
</tr>
</thead>
<tbody>
<tr>
<td>Llama 3 70B Instruct</td>
<td class="has-text-centered">0.715</td>
<td class="has-text-centered">0.693</td>
<td class="has-text-centered">0.701</td>
<td class="has-text-centered">0.911</td>
<td class="has-text-centered">0.314</td>
<td class="has-text-centered performance-medium">0.454</td>
<td class="has-text-centered">0.314</td>
<td class="has-text-centered">0.332</td>
<td class="has-text-centered">0.879</td>
<td class="has-text-centered">0.904</td>
<td class="has-text-centered">0.879</td>
<td class="has-text-centered">0.883</td>
<td class="has-text-centered">0.015</td>
<td class="has-text-centered">0.030</td>
<td class="has-text-centered">0.020</td>
<td class="has-text-centered">0.010</td>
<td class="has-text-centered">0.474</td>
<td class="has-text-centered">0.485</td>
<td class="has-text-centered">0.485</td>
<td class="has-text-centered">0.469</td>
</tr>
<tr>
<td>Llama 3 8B Instruct</td>
<td class="has-text-centered">0.581</td>
<td class="has-text-centered">0.558</td>
<td class="has-text-centered">0.565</td>
<td class="has-text-centered">0.854</td>
<td class="has-text-centered">0.296</td>
<td class="has-text-centered">0.357</td>
<td class="has-text-centered">0.296</td>
<td class="has-text-centered">0.289</td>
<td class="has-text-centered">0.723</td>
<td class="has-text-centered">0.755</td>
<td class="has-text-centered">0.723</td>
<td class="has-text-centered">0.705</td>
<td class="has-text-centered">0.003</td>
<td class="has-text-centered">0.004</td>
<td class="has-text-centered">0.003</td>
<td class="has-text-centered">0.002</td>
<td class="has-text-centered">0.301</td>
<td class="has-text-centered">0.478</td>
<td class="has-text-centered">0.478</td>
<td class="has-text-centered">0.350</td>
</tr>
<tr>
<td>DBRX Instruct</td>
<td class="has-text-centered">0.516</td>
<td class="has-text-centered">0.476</td>
<td class="has-text-centered">0.489</td>
<td class="has-text-centered">0.802</td>
<td class="has-text-centered">0.329</td>
<td class="has-text-centered">0.371</td>
<td class="has-text-centered">0.329</td>
<td class="has-text-centered">0.304</td>
<td class="has-text-centered">0.766</td>
<td class="has-text-centered">0.825</td>
<td class="has-text-centered">0.766</td>
<td class="has-text-centered">0.778</td>
<td class="has-text-centered">0.008</td>
<td class="has-text-centered">0.011</td>
<td class="has-text-centered">0.009</td>
<td class="has-text-centered">0.005</td>
<td class="has-text-centered">0.004</td>
<td class="has-text-centered">0.014</td>
<td class="has-text-centered">0.014</td>
<td class="has-text-centered">0.006</td>
</tr>
<tr>
<td>DeepSeek LLM (67B)</td>
<td class="has-text-centered">0.752</td>
<td class="has-text-centered">0.742</td>
<td class="has-text-centered">0.745</td>
<td class="has-text-centered">0.917</td>
<td class="has-text-centered">0.344</td>
<td class="has-text-centered">0.403</td>
<td class="has-text-centered">0.344</td>
<td class="has-text-centered">0.334</td>
<td class="has-text-centered">0.874</td>
<td class="has-text-centered">0.890</td>
<td class="has-text-centered">0.874</td>
<td class="has-text-centered">0.879</td>
<td class="has-text-centered">0.005</td>
<td class="has-text-centered">0.009</td>
<td class="has-text-centered">0.007</td>
<td class="has-text-centered">0.003</td>
<td class="has-text-centered">0.456</td>
<td class="has-text-centered">0.405</td>
<td class="has-text-centered">0.405</td>
<td class="has-text-centered">0.416</td>
</tr>
<tr>
<td>Gemma 2 27B</td>
<td class="has-text-centered">0.772</td>
<td class="has-text-centered">0.754</td>
<td class="has-text-centered">0.761</td>
<td class="has-text-centered performance-medium">0.923</td>
<td class="has-text-centered">0.352</td>
<td class="has-text-centered">0.437</td>
<td class="has-text-centered">0.352</td>
<td class="has-text-centered">0.356</td>
<td class="has-text-centered">0.897</td>
<td class="has-text-centered">0.914</td>
<td class="has-text-centered">0.897</td>
<td class="has-text-centered">0.902</td>
<td class="has-text-centered">0.005</td>
<td class="has-text-centered">0.008</td>
<td class="has-text-centered">0.006</td>
<td class="has-text-centered">0.003</td>
<td class="has-text-centered">0.320</td>
<td class="has-text-centered">0.295</td>
<td class="has-text-centered">0.295</td>
<td class="has-text-centered">0.298</td>
</tr>
<tr>
<td>Gemma 2 9B</td>
<td class="has-text-centered">0.665</td>
<td class="has-text-centered">0.643</td>
<td class="has-text-centered">0.651</td>
<td class="has-text-centered">0.886</td>
<td class="has-text-centered">0.336</td>
<td class="has-text-centered">0.373</td>
<td class="has-text-centered">0.336</td>
<td class="has-text-centered">0.331</td>
<td class="has-text-centered">0.885</td>
<td class="has-text-centered">0.902</td>
<td class="has-text-centered">0.885</td>
<td class="has-text-centered">0.892</td>
<td class="has-text-centered">0.004</td>
<td class="has-text-centered">0.008</td>
<td class="has-text-centered">0.005</td>
<td class="has-text-centered">0.003</td>
<td class="has-text-centered">0.348</td>
<td class="has-text-centered">0.419</td>
<td class="has-text-centered">0.419</td>
<td class="has-text-centered">0.367</td>
</tr>
<tr>
<td>Mistral (7B) Instruct</td>
<td class="has-text-centered">0.540</td>
<td class="has-text-centered">0.522</td>
<td class="has-text-centered">0.526</td>
<td class="has-text-centered">0.806</td>
<td class="has-text-centered">0.278</td>
<td class="has-text-centered">0.383</td>
<td class="has-text-centered">0.278</td>
<td class="has-text-centered">0.276</td>
<td class="has-text-centered">0.767</td>
<td class="has-text-centered">0.817</td>
<td class="has-text-centered">0.767</td>
<td class="has-text-centered">0.771</td>
<td class="has-text-centered">0.004</td>
<td class="has-text-centered">0.006</td>
<td class="has-text-centered">0.004</td>
<td class="has-text-centered">0.002</td>
<td class="has-text-centered">0.337</td>
<td class="has-text-centered">0.477</td>
<td class="has-text-centered">0.477</td>
<td class="has-text-centered">0.368</td>
</tr>
<tr>
<td>Mixtral-8x22B Instruct</td>
<td class="has-text-centered">0.653</td>
<td class="has-text-centered">0.625</td>
<td class="has-text-centered">0.635</td>
<td class="has-text-centered">0.870</td>
<td class="has-text-centered">0.381</td>
<td class="has-text-centered">0.414</td>
<td class="has-text-centered">0.381</td>
<td class="has-text-centered">0.367</td>
<td class="has-text-centered">0.807</td>
<td class="has-text-centered">0.847</td>
<td class="has-text-centered">0.807</td>
<td class="has-text-centered">0.811</td>
<td class="has-text-centered">0.010</td>
<td class="has-text-centered">0.008</td>
<td class="has-text-centered">0.009</td>
<td class="has-text-centered">0.005</td>
<td class="has-text-centered">0.428</td>
<td class="has-text-centered">0.481</td>
<td class="has-text-centered">0.481</td>
<td class="has-text-centered">0.435</td>
</tr>
<tr>
<td>Mixtral-8x7B Instruct</td>
<td class="has-text-centered">0.613</td>
<td class="has-text-centered">0.591</td>
<td class="has-text-centered">0.598</td>
<td class="has-text-centered">0.875</td>
<td class="has-text-centered">0.291</td>
<td class="has-text-centered">0.376</td>
<td class="has-text-centered">0.291</td>
<td class="has-text-centered">0.282</td>
<td class="has-text-centered">0.840</td>
<td class="has-text-centered">0.863</td>
<td class="has-text-centered">0.840</td>
<td class="has-text-centered">0.845</td>
<td class="has-text-centered">0.007</td>
<td class="has-text-centered">0.012</td>
<td class="has-text-centered">0.009</td>
<td class="has-text-centered">0.005</td>
<td class="has-text-centered">0.251</td>
<td class="has-text-centered">0.324</td>
<td class="has-text-centered">0.324</td>
<td class="has-text-centered">0.267</td>
</tr>
<tr>
<td>Qwen 2 Instruct (72B)</td>
<td class="has-text-centered">0.766</td>
<td class="has-text-centered">0.742</td>
<td class="has-text-centered">0.748</td>
<td class="has-text-centered">0.899</td>
<td class="has-text-centered">0.365</td>
<td class="has-text-centered">0.407</td>
<td class="has-text-centered">0.365</td>
<td class="has-text-centered">0.348</td>
<td class="has-text-centered">0.850</td>
<td class="has-text-centered">0.881</td>
<td class="has-text-centered">0.850</td>
<td class="has-text-centered">0.854</td>
<td class="has-text-centered">0.010</td>
<td class="has-text-centered">0.016</td>
<td class="has-text-centered">0.012</td>
<td class="has-text-centered">0.006</td>
<td class="has-text-centered">0.468</td>
<td class="has-text-centered">0.530</td>
<td class="has-text-centered">0.530</td>
<td class="has-text-centered">0.483</td>
</tr>
<tr>
<td>WizardLM-2 8x22B</td>
<td class="has-text-centered">0.755</td>
<td class="has-text-centered">0.741</td>
<td class="has-text-centered">0.744</td>
<td class="has-text-centered">0.920</td>
<td class="has-text-centered">0.362</td>
<td class="has-text-centered">0.397</td>
<td class="has-text-centered">0.362</td>
<td class="has-text-centered">0.355</td>
<td class="has-text-centered">0.846</td>
<td class="has-text-centered">0.874</td>
<td class="has-text-centered">0.846</td>
<td class="has-text-centered">0.852</td>
<td class="has-text-centered">0.008</td>
<td class="has-text-centered">0.009</td>
<td class="has-text-centered">0.008</td>
<td class="has-text-centered">0.004</td>
<td class="has-text-centered">0.222</td>
<td class="has-text-centered">0.247</td>
<td class="has-text-centered">0.247</td>
<td class="has-text-centered">0.226</td>
</tr>
<tr>
<td>DeepSeek-V3</td>
<td class="has-text-centered performance-low">0.798</td>
<td class="has-text-centered performance-low">0.787</td>
<td class="has-text-centered performance-low">0.790</td>
<td class="has-text-centered performance-best">0.945</td>
<td class="has-text-centered performance-medium">0.450</td>
<td class="has-text-centered performance-medium">0.463</td>
<td class="has-text-centered performance-medium">0.450</td>
<td class="has-text-centered performance-medium">0.437</td>
<td class="has-text-centered">0.927</td>
<td class="has-text-centered performance-low">0.943</td>
<td class="has-text-centered">0.927</td>
<td class="has-text-centered">0.934</td>
<td class="has-text-centered performance-medium">0.034</td>
<td class="has-text-centered performance-low">0.067</td>
<td class="has-text-centered performance-low">0.045</td>
<td class="has-text-centered performance-low">0.023</td>
<td class="has-text-centered">0.563</td>
<td class="has-text-centered">0.544</td>
<td class="has-text-centered">0.544</td>
<td class="has-text-centered">0.549</td>
</tr>
<tr>
<td>DeepSeek R1</td>
<td class="has-text-centered performance-best">0.813</td>
<td class="has-text-centered performance-best">0.805</td>
<td class="has-text-centered performance-best">0.807</td>
<td class="has-text-centered performance-medium">0.944</td>
<td class="has-text-centered performance-low">0.412</td>
<td class="has-text-centered">0.424</td>
<td class="has-text-centered performance-low">0.412</td>
<td class="has-text-centered">0.393</td>
<td class="has-text-centered performance-best">0.946</td>
<td class="has-text-centered performance-best">0.960</td>
<td class="has-text-centered performance-best">0.946</td>
<td class="has-text-centered performance-best">0.952</td>
<td class="has-text-centered performance-best">0.044</td>
<td class="has-text-centered performance-best">0.082</td>
<td class="has-text-centered performance-best">0.057</td>
<td class="has-text-centered performance-best">0.029</td>
<td class="has-text-centered performance-low">0.600</td>
<td class="has-text-centered performance-low">0.586</td>
<td class="has-text-centered performance-low">0.586</td>
<td class="has-text-centered performance-low">0.587</td>
</tr>
<tr>
<td>QwQ-32B-Preview</td>
<td class="has-text-centered">0.695</td>
<td class="has-text-centered">0.681</td>
<td class="has-text-centered">0.685</td>
<td class="has-text-centered">0.907</td>
<td class="has-text-centered">0.278</td>
<td class="has-text-centered">0.396</td>
<td class="has-text-centered">0.278</td>
<td class="has-text-centered">0.270</td>
<td class="has-text-centered">0.680</td>
<td class="has-text-centered">0.770</td>
<td class="has-text-centered">0.680</td>
<td class="has-text-centered">0.656</td>
<td class="has-text-centered">0.001</td>
<td class="has-text-centered">0.001</td>
<td class="has-text-centered">0.001</td>
<td class="has-text-centered">0.000</td>
<td class="has-text-centered">0.005</td>
<td class="has-text-centered">0.005</td>
<td class="has-text-centered">0.005</td>
<td class="has-text-centered">0.005</td>
</tr>
<tr>
<td>Jamba 1.5 Mini</td>
<td class="has-text-centered">0.564</td>
<td class="has-text-centered">0.556</td>
<td class="has-text-centered">0.552</td>
<td class="has-text-centered">0.818</td>
<td class="has-text-centered">0.308</td>
<td class="has-text-centered">0.450</td>
<td class="has-text-centered">0.308</td>
<td class="has-text-centered">0.284</td>
<td class="has-text-centered">0.830</td>
<td class="has-text-centered">0.864</td>
<td class="has-text-centered">0.830</td>
<td class="has-text-centered">0.844</td>
<td class="has-text-centered">0.004</td>
<td class="has-text-centered">0.006</td>
<td class="has-text-centered">0.005</td>
<td class="has-text-centered">0.003</td>
<td class="has-text-centered">0.119</td>
<td class="has-text-centered">0.182</td>
<td class="has-text-centered">0.182</td>
<td class="has-text-centered">0.132</td>
</tr>
<tr>
<td>Jamba 1.5 Large</td>
<td class="has-text-centered">0.707</td>
<td class="has-text-centered">0.687</td>
<td class="has-text-centered">0.693</td>
<td class="has-text-centered">0.883</td>
<td class="has-text-centered">0.341</td>
<td class="has-text-centered">0.452</td>
<td class="has-text-centered">0.341</td>
<td class="has-text-centered">0.341</td>
<td class="has-text-centered">0.856</td>
<td class="has-text-centered">0.890</td>
<td class="has-text-centered">0.856</td>
<td class="has-text-centered">0.862</td>
<td class="has-text-centered">0.004</td>
<td class="has-text-centered">0.005</td>
<td class="has-text-centered">0.005</td>
<td class="has-text-centered">0.002</td>
<td class="has-text-centered">0.403</td>
<td class="has-text-centered">0.414</td>
<td class="has-text-centered">0.414</td>
<td class="has-text-centered">0.397</td>
</tr>
<tr>
<td>Claude 3.5 Sonnet</td>
<td class="has-text-centered performance-medium">0.811</td>
<td class="has-text-centered performance-medium">0.794</td>
<td class="has-text-centered performance-medium">0.799</td>
<td class="has-text-centered">0.922</td>
<td class="has-text-centered performance-best">0.455</td>
<td class="has-text-centered performance-best">0.465</td>
<td class="has-text-centered performance-best">0.455</td>
<td class="has-text-centered performance-best">0.439</td>
<td class="has-text-centered">0.873</td>
<td class="has-text-centered">0.927</td>
<td class="has-text-centered">0.873</td>
<td class="has-text-centered">0.891</td>
<td class="has-text-centered performance-medium">0.034</td>
<td class="has-text-centered performance-medium">0.080</td>
<td class="has-text-centered performance-medium">0.047</td>
<td class="has-text-centered performance-medium">0.024</td>
<td class="has-text-centered performance-medium">0.658</td>
<td class="has-text-centered performance-medium">0.668</td>
<td class="has-text-centered performance-medium">0.668</td>
<td class="has-text-centered performance-medium">0.655</td>
</tr>
<tr>
<td>Claude 3 Haiku</td>
<td class="has-text-centered">0.732</td>
<td class="has-text-centered">0.700</td>
<td class="has-text-centered">0.711</td>
<td class="has-text-centered">0.895</td>
<td class="has-text-centered">0.294</td>
<td class="has-text-centered">0.330</td>
<td class="has-text-centered">0.294</td>
<td class="has-text-centered">0.285</td>
<td class="has-text-centered">0.879</td>
<td class="has-text-centered">0.917</td>
<td class="has-text-centered">0.879</td>
<td class="has-text-centered">0.883</td>
<td class="has-text-centered">0.011</td>
<td class="has-text-centered">0.022</td>
<td class="has-text-centered">0.015</td>
<td class="has-text-centered">0.008</td>
<td class="has-text-centered">0.498</td>
<td class="has-text-centered">0.517</td>
<td class="has-text-centered">0.517</td>
<td class="has-text-centered">0.494</td>
</tr>
<tr>
<td>Cohere Command R +</td>
<td class="has-text-centered">0.769</td>
<td class="has-text-centered">0.750</td>
<td class="has-text-centered">0.756</td>
<td class="has-text-centered">0.902</td>
<td class="has-text-centered">0.353</td>
<td class="has-text-centered">0.405</td>
<td class="has-text-centered">0.353</td>
<td class="has-text-centered">0.333</td>
<td class="has-text-centered">0.917</td>
<td class="has-text-centered">0.930</td>
<td class="has-text-centered">0.917</td>
<td class="has-text-centered">0.922</td>
<td class="has-text-centered">0.016</td>
<td class="has-text-centered">0.032</td>
<td class="has-text-centered">0.021</td>
<td class="has-text-centered">0.011</td>
<td class="has-text-centered">0.462</td>
<td class="has-text-centered">0.459</td>
<td class="has-text-centered">0.459</td>
<td class="has-text-centered">0.452</td>
</tr>
<tr>
<td>Google Gemini 1.5 Pro</td>
<td class="has-text-centered">0.728</td>
<td class="has-text-centered">0.705</td>
<td class="has-text-centered">0.712</td>
<td class="has-text-centered">0.891</td>
<td class="has-text-centered">0.373</td>
<td class="has-text-centered">0.436</td>
<td class="has-text-centered">0.373</td>
<td class="has-text-centered">0.374</td>
<td class="has-text-centered performance-medium">0.934</td>
<td class="has-text-centered performance-medium">0.955</td>
<td class="has-text-centered performance-medium">0.934</td>
<td class="has-text-centered performance-medium">0.944</td>
<td class="has-text-centered">0.014</td>
<td class="has-text-centered">0.028</td>
<td class="has-text-centered">0.019</td>
<td class="has-text-centered">0.010</td>
<td class="has-text-centered">0.399</td>
<td class="has-text-centered">0.400</td>
<td class="has-text-centered">0.400</td>
<td class="has-text-centered">0.393</td>
</tr>
<tr>
<td>OpenAI gpt-4o</td>
<td class="has-text-centered">0.778</td>
<td class="has-text-centered">0.760</td>
<td class="has-text-centered">0.766</td>
<td class="has-text-centered">0.911</td>
<td class="has-text-centered">0.402</td>
<td class="has-text-centered">0.445</td>
<td class="has-text-centered">0.402</td>
<td class="has-text-centered">0.399</td>
<td class="has-text-centered performance-low">0.931</td>
<td class="has-text-centered performance-medium">0.955</td>
<td class="has-text-centered performance-low">0.931</td>
<td class="has-text-centered performance-low">0.942</td>
<td class="has-text-centered performance-low">0.027</td>
<td class="has-text-centered">0.056</td>
<td class="has-text-centered">0.037</td>
<td class="has-text-centered">0.019</td>
<td class="has-text-centered">0.537</td>
<td class="has-text-centered">0.517</td>
<td class="has-text-centered">0.517</td>
<td class="has-text-centered">0.523</td>
</tr>
<tr>
<td>OpenAI o1-mini</td>
<td class="has-text-centered">0.772</td>
<td class="has-text-centered">0.755</td>
<td class="has-text-centered">0.761</td>
<td class="has-text-centered">0.922</td>
<td class="has-text-centered">0.407</td>
<td class="has-text-centered">0.444</td>
<td class="has-text-centered">0.407</td>
<td class="has-text-centered performance-low">0.403</td>
<td class="has-text-centered">0.867</td>
<td class="has-text-centered">0.900</td>
<td class="has-text-centered">0.867</td>
<td class="has-text-centered">0.876</td>
<td class="has-text-centered">0.007</td>
<td class="has-text-centered">0.015</td>
<td class="has-text-centered">0.010</td>
<td class="has-text-centered">0.005</td>
<td class="has-text-centered performance-best">0.661</td>
<td class="has-text-centered performance-best">0.681</td>
<td class="has-text-centered performance-best">0.681</td>
<td class="has-text-centered performance-best">0.662</td>
</tr>
</tbody>
</table>
<div class="content is-small mt-4">
<p><strong>Note:</strong> Color highlighting indicates performance ranking:
<span class="performance-best"> Best </span>,
<span class="performance-medium"> Strong </span>,
<span class="performance-low"> Good </span>
</p>
</div>
</div>
</div>
<!-- Question Answering tab content -->
<div id="question-answering" class="tab-content">
<h2 class="title is-4">Question Answering Task Results</h2>
<div class="results-table">
<table class="table is-bordered is-striped is-narrow is-hoverable is-fullwidth">
<thead>
<tr>
<th rowspan="2">Model</th>
<th colspan="3" class="has-text-centered">Datasets (Accuracy)</th>
</tr>
<tr>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="FinQA" data-tooltip="FinQA contains 8,281 question-answer pairs derived from financial reports that require numerical reasoning over tabular financial data. The question-answering task features multi-step reasoning challenges with full annotation of reasoning programs to solve complex financial queries.">FinQA</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="ConvFinQA" data-tooltip="ConvFinQA is a multi-turn question answering dataset with 3,892 conversations containing 14,115 questions that explore chains of numerical reasoning in financial contexts. The conversational task requires maintaining context while performing sequential numerical operations to answer increasingly complex financial questions.">ConvFinQA</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="TATQA" data-tooltip="TATQA is a large-scale question answering dataset for hybrid data sources that combines tables and text from financial reports. The task emphasizes numerical reasoning operations across multiple formats, requiring models to integrate information from structured and unstructured sources to answer financial questions.">TATQA</th>
</tr>
</thead>
<tbody>
<tr>
<td>Llama 3 70B Instruct</td>
<td class="has-text-centered">0.809</td>
<td class="has-text-centered">0.709</td>
<td class="has-text-centered">0.772</td>
</tr>
<tr>
<td>Llama 3 8B Instruct</td>
<td class="has-text-centered">0.767</td>
<td class="has-text-centered">0.268</td>
<td class="has-text-centered">0.706</td>
</tr>
<tr>
<td>DBRX Instruct</td>
<td class="has-text-centered">0.738</td>
<td class="has-text-centered">0.252</td>
<td class="has-text-centered">0.633</td>
</tr>
<tr>
<td>DeepSeek LLM (67B)</td>
<td class="has-text-centered">0.742</td>
<td class="has-text-centered">0.174</td>
<td class="has-text-centered">0.355</td>
</tr>
<tr>
<td>Gemma 2 27B</td>
<td class="has-text-centered">0.768</td>
<td class="has-text-centered">0.268</td>
<td class="has-text-centered">0.734</td>
</tr>
<tr>
<td>Gemma 2 9B</td>
<td class="has-text-centered">0.779</td>
<td class="has-text-centered">0.292</td>
<td class="has-text-centered">0.750</td>
</tr>
<tr>
<td>Mistral (7B) Instruct v0.3</td>
<td class="has-text-centered">0.655</td>
<td class="has-text-centered">0.199</td>
<td class="has-text-centered">0.553</td>
</tr>
<tr>
<td>Mixtral-8x22B Instruct</td>
<td class="has-text-centered">0.766</td>
<td class="has-text-centered">0.285</td>
<td class="has-text-centered">0.666</td>
</tr>
<tr>
<td>Mixtral-8x7B Instruct</td>
<td class="has-text-centered">0.611</td>
<td class="has-text-centered">0.315</td>
<td class="has-text-centered">0.501</td>
</tr>
<tr>
<td>Qwen 2 Instruct (72B)</td>
<td class="has-text-centered">0.819</td>
<td class="has-text-centered">0.269</td>
<td class="has-text-centered">0.715</td>
</tr>
<tr>
<td>WizardLM-2 8x22B</td>
<td class="has-text-centered">0.796</td>
<td class="has-text-centered">0.247</td>
<td class="has-text-centered">0.725</td>
</tr>
<tr>
<td>DeepSeek-V3</td>
<td class="has-text-centered performance-medium">0.840</td>
<td class="has-text-centered">0.261</td>
<td class="has-text-centered performance-low">0.779</td>
</tr>
<tr>
<td>DeepSeek R1</td>
<td class="has-text-centered performance-low">0.836</td>
<td class="has-text-centered performance-best">0.853</td>
<td class="has-text-centered performance-best">0.858</td>
</tr>
<tr>
<td>QwQ-32B-Preview</td>
<td class="has-text-centered">0.793</td>
<td class="has-text-centered">0.282</td>
<td class="has-text-centered performance-medium">0.796</td>
</tr>
<tr>
<td>Jamba 1.5 Mini</td>
<td class="has-text-centered">0.666</td>
<td class="has-text-centered">0.218</td>
<td class="has-text-centered">0.586</td>
</tr>
<tr>
<td>Jamba 1.5 Large</td>
<td class="has-text-centered">0.790</td>
<td class="has-text-centered">0.225</td>
<td class="has-text-centered">0.660</td>
</tr>
<tr>
<td>Claude 3.5 Sonnet</td>
<td class="has-text-centered performance-best">0.844</td>
<td class="has-text-centered">0.402</td>
<td class="has-text-centered">0.700</td>
</tr>
<tr>
<td>Claude 3 Haiku</td>
<td class="has-text-centered">0.803</td>
<td class="has-text-centered">0.421</td>
<td class="has-text-centered">0.733</td>
</tr>
<tr>
<td>Cohere Command R 7B</td>
<td class="has-text-centered">0.709</td>
<td class="has-text-centered">0.212</td>
<td class="has-text-centered">0.716</td>
</tr>
<tr>
<td>Cohere Command R +</td>
<td class="has-text-centered">0.776</td>
<td class="has-text-centered">0.259</td>
<td class="has-text-centered">0.698</td>
</tr>
<tr>
<td>Google Gemini 1.5 Pro</td>
<td class="has-text-centered">0.829</td>
<td class="has-text-centered">0.280</td>
<td class="has-text-centered">0.763</td>
</tr>
<tr>
<td>OpenAI gpt-4o</td>
<td class="has-text-centered performance-low">0.836</td>
<td class="has-text-centered performance-low">0.749</td>
<td class="has-text-centered">0.754</td>
</tr>
<tr>
<td>OpenAI o1-mini</td>
<td class="has-text-centered">0.799</td>
<td class="has-text-centered performance-medium">0.840</td>
<td class="has-text-centered">0.698</td>
</tr>
</tbody>
</table>
<div class="content is-small mt-4">
<p><strong>Note:</strong> Color highlighting indicates performance ranking:
<span class="performance-best"> Best </span>,
<span class="performance-medium"> Strong </span>,
<span class="performance-low"> Good </span>
</p>
</div>
</div>
</div>
<!-- Sentiment Analysis tab content -->
<div id="sentiment-analysis" class="tab-content">
<h2 class="title is-4">Sentiment Analysis Task Results</h2>
<div class="results-table">
<table class="table is-bordered is-striped is-narrow is-hoverable is-fullwidth">
<thead>
<tr>
<th rowspan="2">Model</th>
<th colspan="3" class="has-text-centered tooltip-trigger tooltip-right" data-title="FiQA Task 1" data-tooltip="FiQA Task 1 focuses on aspect-based financial sentiment analysis in microblog posts and news headlines using a continuous scale from -1 (negative) to 1 (positive). The regression task requires models to accurately predict the sentiment score that reflects investor perception of financial texts.">FiQA Task 1</th>
<th colspan="4" class="has-text-centered tooltip-trigger tooltip-right" data-table="FinEntity" data-tooltip="FinEntity contains 1,000 financial news articles annotated with sentiment labels for 10 financial entities. The multi-label classification task involves predicting the sentiment of each entity in the news article, requiring models to understand the nuanced sentiment expressed towards different financial entities.">FinEntity</th>
<th colspan="4" class="has-text-centered tooltip-trigger tooltip-right" data-title="SubjECTive-QA" data-tooltip="SubjECTive-QA contains 49,446 annotations across 2,747 question-answer pairs extracted from 120 earnings call transcripts. The multi-label classification task involves analyzing six subjective features in financial discourse: assertiveness, cautiousness, optimism, specificity, clarity, and relevance.">SubjECTive-QA</th>
<th colspan="4" class="has-text-centered tooltip-trigger tooltip-right" data-title="Financial Phrase Bank" data-tooltip="Financial Phrase Bank (FPB) contains 4,840 sentences from financial news articles categorized as positive, negative, or neutral by 16 finance experts using majority voting. The sentiment classification task requires understanding how these statements might influence investor perception of stock prices.">Financial Phrase Bank (FPB)</th>
</tr>
<tr>
<th class="has-text-centered">MSE</th>
<th class="has-text-centered">MAE</th>
<th class="has-text-centered">r² Score</th>
<th class="has-text-centered">Precision</th>
<th class="has-text-centered">Recall</th>
<th class="has-text-centered">Accuracy</th>
<th class="has-text-centered">F1</th>
<th class="has-text-centered">Precision</th>
<th class="has-text-centered">Recall</th>
<th class="has-text-centered">F1</th>
<th class="has-text-centered">Accuracy</th>
<th class="has-text-centered">Accuracy</th>
<th class="has-text-centered">Precision</th>
<th class="has-text-centered">Recall</th>
<th class="has-text-centered">F1</th>
</tr>
</thead>
<tbody>
<tr>
<td>Llama 3 70B Instruct</td>
<td class="has-text-centered">0.123</td>
<td class="has-text-centered">0.290</td>
<td class="has-text-centered">0.272</td>
<td class="has-text-centered">0.474</td>
<td class="has-text-centered">0.485</td>
<td class="has-text-centered">0.485</td>
<td class="has-text-centered">0.469</td>
<td class="has-text-centered">0.652</td>
<td class="has-text-centered">0.573</td>
<td class="has-text-centered">0.535</td>
<td class="has-text-centered">0.573</td>
<td class="has-text-centered">0.901</td>
<td class="has-text-centered">0.904</td>
<td class="has-text-centered">0.901</td>
<td class="has-text-centered">0.902</td>
</tr>
<tr>
<td>Llama 3 8B Instruct</td>
<td class="has-text-centered">0.161</td>
<td class="has-text-centered">0.344</td>
<td class="has-text-centered">0.045</td>
<td class="has-text-centered">0.301</td>
<td class="has-text-centered">0.478</td>
<td class="has-text-centered">0.478</td>
<td class="has-text-centered">0.350</td>
<td class="has-text-centered">0.635</td>
<td class="has-text-centered performance-best">0.625</td>
<td class="has-text-centered performance-best">0.600</td>
<td class="has-text-centered performance-best">0.625</td>
<td class="has-text-centered">0.738</td>
<td class="has-text-centered">0.801</td>
<td class="has-text-centered">0.738</td>
<td class="has-text-centered">0.698</td>
</tr>
<tr>
<td>DBRX Instruct</td>
<td class="has-text-centered">0.160</td>
<td class="has-text-centered">0.321</td>
<td class="has-text-centered">0.052</td>
<td class="has-text-centered">0.004</td>
<td class="has-text-centered">0.014</td>
<td class="has-text-centered">0.014</td>
<td class="has-text-centered">0.006</td>
<td class="has-text-centered performance-low">0.654</td>
<td class="has-text-centered">0.541</td>
<td class="has-text-centered">0.436</td>
<td class="has-text-centered">0.541</td>
<td class="has-text-centered">0.524</td>
<td class="has-text-centered">0.727</td>
<td class="has-text-centered">0.524</td>
<td class="has-text-centered">0.499</td>
</tr>
<tr>
<td>DeepSeek LLM (67B)</td>
<td class="has-text-centered">0.118</td>
<td class="has-text-centered">0.278</td>
<td class="has-text-centered">0.302</td>
<td class="has-text-centered">0.456</td>
<td class="has-text-centered">0.405</td>
<td class="has-text-centered">0.405</td>
<td class="has-text-centered">0.416</td>
<td class="has-text-centered performance-best">0.676</td>
<td class="has-text-centered">0.544</td>
<td class="has-text-centered">0.462</td>
<td class="has-text-centered">0.544</td>
<td class="has-text-centered">0.815</td>
<td class="has-text-centered">0.867</td>
<td class="has-text-centered">0.815</td>
<td class="has-text-centered">0.811</td>
</tr>
<tr>
<td>Gemma 2 27B</td>
<td class="has-text-centered performance-best">0.100</td>
<td class="has-text-centered performance-best">0.266</td>
<td class="has-text-centered">0.406</td>
<td class="has-text-centered">0.320</td>
<td class="has-text-centered">0.295</td>
<td class="has-text-centered">0.295</td>
<td class="has-text-centered">0.298</td>
<td class="has-text-centered">0.562</td>
<td class="has-text-centered">0.524</td>
<td class="has-text-centered">0.515</td>
<td class="has-text-centered">0.524</td>
<td class="has-text-centered">0.890</td>
<td class="has-text-centered">0.896</td>
<td class="has-text-centered">0.890</td>
<td class="has-text-centered">0.884</td>
</tr>
<tr>
<td>Gemma 2 9B</td>
<td class="has-text-centered">0.189</td>
<td class="has-text-centered">0.352</td>
<td class="has-text-centered">-0.120</td>
<td class="has-text-centered">0.348</td>
<td class="has-text-centered">0.419</td>
<td class="has-text-centered">0.419</td>
<td class="has-text-centered">0.367</td>
<td class="has-text-centered">0.570</td>
<td class="has-text-centered">0.499</td>
<td class="has-text-centered">0.491</td>
<td class="has-text-centered">0.499</td>
<td class="has-text-centered performance-medium">0.940</td>
<td class="has-text-centered performance-medium">0.941</td>
<td class="has-text-centered performance-medium">0.940</td>
<td class="has-text-centered performance-medium">0.940</td>
</tr>
<tr>
<td>Mistral (7B) Instruct v0.3</td>
<td class="has-text-centered">0.135</td>
<td class="has-text-centered">0.278</td>
<td class="has-text-centered">0.200</td>
<td class="has-text-centered">0.337</td>
<td class="has-text-centered">0.477</td>
<td class="has-text-centered">0.477</td>
<td class="has-text-centered">0.368</td>
<td class="has-text-centered">0.607</td>
<td class="has-text-centered">0.542</td>
<td class="has-text-centered">0.522</td>
<td class="has-text-centered">0.542</td>
<td class="has-text-centered">0.847</td>
<td class="has-text-centered">0.854</td>
<td class="has-text-centered">0.847</td>
<td class="has-text-centered">0.841</td>
</tr>
<tr>
<td>Mixtral-8x22B Instruct</td>
<td class="has-text-centered">0.221</td>
<td class="has-text-centered">0.364</td>
<td class="has-text-centered performance-best">-0.310</td>
<td class="has-text-centered">0.428</td>
<td class="has-text-centered">0.481</td>
<td class="has-text-centered">0.481</td>
<td class="has-text-centered">0.435</td>
<td class="has-text-centered">0.614</td>
<td class="has-text-centered">0.538</td>
<td class="has-text-centered">0.510</td>
<td class="has-text-centered">0.538</td>
<td class="has-text-centered">0.768</td>
<td class="has-text-centered">0.845</td>
<td class="has-text-centered">0.768</td>
<td class="has-text-centered">0.776</td>
</tr>
<tr>
<td>Mixtral-8x7B Instruct</td>
<td class="has-text-centered">0.208</td>
<td class="has-text-centered">0.307</td>
<td class="has-text-centered performance-medium">-0.229</td>
<td class="has-text-centered">0.251</td>
<td class="has-text-centered">0.324</td>
<td class="has-text-centered">0.324</td>
<td class="has-text-centered">0.267</td>
<td class="has-text-centered">0.611</td>
<td class="has-text-centered">0.518</td>
<td class="has-text-centered">0.498</td>
<td class="has-text-centered">0.518</td>
<td class="has-text-centered">0.896</td>
<td class="has-text-centered">0.898</td>
<td class="has-text-centered">0.896</td>
<td class="has-text-centered">0.893</td>
</tr>
<tr>
<td>Qwen 2 Instruct (72B)</td>
<td class="has-text-centered">0.205</td>
<td class="has-text-centered">0.409</td>
<td class="has-text-centered performance-low">-0.212</td>
<td class="has-text-centered">0.468</td>
<td class="has-text-centered">0.530</td>
<td class="has-text-centered">0.530</td>
<td class="has-text-centered">0.483</td>
<td class="has-text-centered">0.644</td>
<td class="has-text-centered performance-medium">0.601</td>
<td class="has-text-centered">0.576</td>
<td class="has-text-centered performance-medium">0.601</td>
<td class="has-text-centered">0.904</td>
<td class="has-text-centered">0.908</td>
<td class="has-text-centered">0.904</td>
<td class="has-text-centered">0.901</td>
</tr>
<tr>
<td>WizardLM-2 8x22B</td>
<td class="has-text-centered">0.129</td>
<td class="has-text-centered">0.283</td>
<td class="has-text-centered">0.239</td>
<td class="has-text-centered">0.222</td>
<td class="has-text-centered">0.247</td>
<td class="has-text-centered">0.247</td>
<td class="has-text-centered">0.226</td>
<td class="has-text-centered">0.611</td>
<td class="has-text-centered">0.570</td>
<td class="has-text-centered">0.566</td>
<td class="has-text-centered">0.570</td>
<td class="has-text-centered">0.765</td>
<td class="has-text-centered">0.853</td>
<td class="has-text-centered">0.765</td>
<td class="has-text-centered">0.779</td>
</tr>
<tr>
<td>DeepSeek-V3</td>
<td class="has-text-centered">0.150</td>
<td class="has-text-centered">0.311</td>
<td class="has-text-centered">0.111</td>
<td class="has-text-centered">0.563</td>
<td class="has-text-centered">0.544</td>
<td class="has-text-centered">0.544</td>
<td class="has-text-centered">0.549</td>
<td class="has-text-centered">0.640</td>
<td class="has-text-centered">0.572</td>
<td class="has-text-centered performance-low">0.583</td>
<td class="has-text-centered">0.572</td>
<td class="has-text-centered">0.828</td>
<td class="has-text-centered">0.851</td>
<td class="has-text-centered">0.828</td>
<td class="has-text-centered">0.814</td>
</tr>
<tr>
<td>DeepSeek R1</td>
<td class="has-text-centered">0.110</td>
<td class="has-text-centered">0.289</td>
<td class="has-text-centered">0.348</td>
<td class="has-text-centered performance-low">0.600</td>
<td class="has-text-centered performance-low">0.586</td>
<td class="has-text-centered performance-low">0.586</td>
<td class="has-text-centered performance-low">0.587</td>
<td class="has-text-centered">0.644</td>
<td class="has-text-centered">0.489</td>
<td class="has-text-centered">0.499</td>
<td class="has-text-centered">0.489</td>
<td class="has-text-centered">0.904</td>
<td class="has-text-centered">0.907</td>
<td class="has-text-centered">0.904</td>
<td class="has-text-centered">0.902</td>
</tr>
<tr>
<td>QwQ-32B-Preview</td>
<td class="has-text-centered">0.141</td>
<td class="has-text-centered">0.290</td>
<td class="has-text-centered">0.165</td>
<td class="has-text-centered">0.005</td>
<td class="has-text-centered">0.005</td>
<td class="has-text-centered">0.005</td>
<td class="has-text-centered">0.005</td>
<td class="has-text-centered">0.629</td>
<td class="has-text-centered">0.534</td>
<td class="has-text-centered">0.550</td>
<td class="has-text-centered">0.534</td>
<td class="has-text-centered">0.812</td>
<td class="has-text-centered">0.827</td>
<td class="has-text-centered">0.812</td>
<td class="has-text-centered">0.815</td>
</tr>
<tr>
<td>Jamba 1.5 Mini</td>
<td class="has-text-centered">0.119</td>
<td class="has-text-centered">0.282</td>
<td class="has-text-centered">0.293</td>
<td class="has-text-centered">0.119</td>
<td class="has-text-centered">0.182</td>
<td class="has-text-centered">0.182</td>
<td class="has-text-centered">0.132</td>
<td class="has-text-centered">0.380</td>
<td class="has-text-centered">0.525</td>
<td class="has-text-centered">0.418</td>
<td class="has-text-centered">0.525</td>
<td class="has-text-centered">0.784</td>
<td class="has-text-centered">0.814</td>
<td class="has-text-centered">0.784</td>
<td class="has-text-centered">0.765</td>
</tr>
<tr>
<td>Jamba 1.5 Large</td>
<td class="has-text-centered">0.183</td>
<td class="has-text-centered">0.363</td>
<td class="has-text-centered">-0.085</td>
<td class="has-text-centered">0.403</td>
<td class="has-text-centered">0.414</td>
<td class="has-text-centered">0.414</td>
<td class="has-text-centered">0.397</td>
<td class="has-text-centered">0.635</td>
<td class="has-text-centered">0.573</td>
<td class="has-text-centered">0.582</td>
<td class="has-text-centered">0.573</td>
<td class="has-text-centered">0.824</td>
<td class="has-text-centered">0.850</td>
<td class="has-text-centered">0.824</td>
<td class="has-text-centered">0.798</td>
</tr>
<tr>
<td>Claude 3.5 Sonnet</td>
<td class="has-text-centered performance-medium">0.101</td>
<td class="has-text-centered performance-medium">0.268</td>
<td class="has-text-centered">0.402</td>
<td class="has-text-centered performance-medium">0.658</td>
<td class="has-text-centered performance-medium">0.668</td>
<td class="has-text-centered performance-medium">0.668</td>
<td class="has-text-centered performance-medium">0.655</td>
<td class="has-text-centered">0.634</td>
<td class="has-text-centered">0.585</td>
<td class="has-text-centered">0.553</td>
<td class="has-text-centered">0.585</td>
<td class="has-text-centered performance-best">0.944</td>
<td class="has-text-centered performance-best">0.945</td>
<td class="has-text-centered performance-best">0.944</td>
<td class="has-text-centered performance-best">0.944</td>
</tr>
<tr>
<td>Claude 3 Haiku</td>
<td class="has-text-centered">0.167</td>
<td class="has-text-centered">0.349</td>
<td class="has-text-centered">0.008</td>
<td class="has-text-centered">0.498</td>
<td class="has-text-centered">0.517</td>
<td class="has-text-centered">0.517</td>
<td class="has-text-centered">0.494</td>
<td class="has-text-centered">0.619</td>
<td class="has-text-centered">0.538</td>
<td class="has-text-centered">0.463</td>
<td class="has-text-centered">0.538</td>
<td class="has-text-centered">0.907</td>
<td class="has-text-centered">0.913</td>
<td class="has-text-centered">0.907</td>
<td class="has-text-centered">0.908</td>
</tr>
<tr>
<td>Cohere Command R 7B</td>
<td class="has-text-centered">0.164</td>
<td class="has-text-centered">0.319</td>
<td class="has-text-centered">0.028</td>
<td class="has-text-centered">0.457</td>
<td class="has-text-centered">0.446</td>
<td class="has-text-centered">0.446</td>
<td class="has-text-centered">0.441</td>
<td class="has-text-centered">0.609</td>
<td class="has-text-centered">0.547</td>
<td class="has-text-centered">0.532</td>
<td class="has-text-centered">0.547</td>
<td class="has-text-centered">0.835</td>
<td class="has-text-centered">0.861</td>
<td class="has-text-centered">0.835</td>
<td class="has-text-centered">0.840</td>
</tr>
<tr>
<td>Cohere Command R +</td>
<td class="has-text-centered performance-low">0.106</td>
<td class="has-text-centered performance-low">0.274</td>
<td class="has-text-centered">0.373</td>
<td class="has-text-centered">0.462</td>
<td class="has-text-centered">0.459</td>
<td class="has-text-centered">0.459</td>
<td class="has-text-centered">0.452</td>
<td class="has-text-centered">0.608</td>
<td class="has-text-centered">0.547</td>
<td class="has-text-centered">0.533</td>
<td class="has-text-centered">0.547</td>
<td class="has-text-centered">0.741</td>
<td class="has-text-centered">0.806</td>
<td class="has-text-centered">0.741</td>
<td class="has-text-centered">0.699</td>
</tr>
<tr>
<td>Google Gemini 1.5 Pro</td>
<td class="has-text-centered">0.144</td>
<td class="has-text-centered">0.329</td>
<td class="has-text-centered">0.149</td>
<td class="has-text-centered">0.399</td>
<td class="has-text-centered">0.400</td>
<td class="has-text-centered">0.400</td>
<td class="has-text-centered">0.393</td>
<td class="has-text-centered">0.642</td>
<td class="has-text-centered performance-low">0.587</td>
<td class="has-text-centered performance-medium">0.593</td>
<td class="has-text-centered performance-low">0.587</td>
<td class="has-text-centered">0.890</td>
<td class="has-text-centered">0.895</td>
<td class="has-text-centered">0.890</td>
<td class="has-text-centered">0.885</td>
</tr>
<tr>
<td>OpenAI gpt-4o</td>
<td class="has-text-centered">0.184</td>
<td class="has-text-centered">0.317</td>
<td class="has-text-centered">-0.089</td>
<td class="has-text-centered">0.537</td>
<td class="has-text-centered">0.517</td>
<td class="has-text-centered">0.517</td>
<td class="has-text-centered">0.523</td>
<td class="has-text-centered">0.639</td>
<td class="has-text-centered">0.515</td>
<td class="has-text-centered">0.541</td>
<td class="has-text-centered">0.515</td>
<td class="has-text-centered performance-low">0.929</td>
<td class="has-text-centered performance-low">0.931</td>
<td class="has-text-centered performance-low">0.929</td>
<td class="has-text-centered performance-low">0.928</td>
</tr>
<tr>
<td>OpenAI o1-mini</td>
<td class="has-text-centered">0.120</td>
<td class="has-text-centered">0.295</td>
<td class="has-text-centered">0.289</td>
<td class="has-text-centered performance-best">0.661</td>
<td class="has-text-centered performance-best">0.681</td>
<td class="has-text-centered performance-best">0.681</td>
<td class="has-text-centered performance-best">0.662</td>
<td class="has-text-centered performance-medium">0.660</td>
<td class="has-text-centered">0.515</td>
<td class="has-text-centered">0.542</td>
<td class="has-text-centered">0.515</td>
<td class="has-text-centered">0.918</td>
<td class="has-text-centered">0.917</td>
<td class="has-text-centered">0.918</td>
<td class="has-text-centered">0.917</td>
</tr>
</tbody>
</table>
<div class="content is-small mt-4">
<p><strong>Note:</strong> Color highlighting indicates performance ranking:
<span class="performance-best"> Best </span>,
<span class="performance-medium"> Strong </span>,
<span class="performance-low"> Good </span>
</p>
</div>
</div>
</div>
<!-- Text Classification tab content -->
<div id="text-classification" class="tab-content">
<h2 class="title is-4">Text Classification Task Results</h2>
<div class="results-table">
<table class="table is-bordered is-striped is-narrow is-hoverable is-fullwidth">
<thead>
<tr>
<th rowspan="2">Model</th>
<th colspan="4" class="has-text-centered tooltip-trigger tooltip-right" data-title="Banking77" data-tooltip="Banking77 is a fine-grained dataset comprising 13,083 customer service queries annotated with 77 unique intents from the banking domain. The task involves accurately classifying each customer query into the correct intent category to improve automated banking support systems.">Banking77</th>
<th colspan="4" class="has-text-centered tooltip-trigger tooltip-right" data-title="FinBench" data-tooltip="FinBench is a comprehensive evaluation dataset containing 333,000 labeled instances that combines tabular data and profile text for financial risk prediction. The task requires models to predict financial outcomes across three key risk categories: default, fraud, and customer churn.">FinBench</th>
<th colspan="4" class="has-text-centered tooltip-trigger tooltip-right" data-title="FOMC" data-tooltip="FOMC is a dataset containing Federal Open Market Committee speeches, meeting minutes, and press conference transcripts spanning from 1996 to 2022. The classification task involves determining whether the monetary policy stance expressed in each document is hawkish (tighter monetary policy) or dovish (looser monetary policy).">FOMC</th>
<th colspan="4" class="has-text-centered tooltip-trigger tooltip-right" data-title="NumClaim" data-tooltip="NumClaim is an expert-annotated dataset for detecting and analyzing fine-grained investor claims within financial narratives that contain numerical information. The task requires identifying and categorizing claims containing numerals in analyst reports and earnings call transcripts for investment decision making.">NumClaim</th>
<th colspan="1" class="has-text-centered tooltip-trigger tooltip-right" data-title="Headlines" data-tooltip="Headlines is a dataset containing 11,412 human-annotated financial news headlines focused on commodities, particularly gold, spanning from 2000 to 2019. The classification task involves identifying binary indicators for price mentions and directional price movements in these concise financial texts.">Headlines</th>
</tr>
<tr>
<th class="has-text-centered">Accuracy</th>
<th class="has-text-centered">Precision</th>
<th class="has-text-centered">Recall</th>
<th class="has-text-centered">F1</th>
<th class="has-text-centered">Accuracy</th>
<th class="has-text-centered">Precision</th>
<th class="has-text-centered">Recall</th>
<th class="has-text-centered">F1</th>
<th class="has-text-centered">Accuracy</th>
<th class="has-text-centered">Precision</th>
<th class="has-text-centered">Recall</th>
<th class="has-text-centered">F1</th>
<th class="has-text-centered">Precision</th>
<th class="has-text-centered">Recall</th>
<th class="has-text-centered">Accuracy</th>
<th class="has-text-centered">F1</th>
<th class="has-text-centered">Accuracy</th>
</tr>
</thead>
<tbody>
<tr>
<td>Llama 3 70B Instruct</td>
<td class="has-text-centered">0.660</td>
<td class="has-text-centered">0.748</td>
<td class="has-text-centered">0.660</td>
<td class="has-text-centered">0.645</td>
<td class="has-text-centered">0.222</td>
<td class="has-text-centered">0.826</td>
<td class="has-text-centered">0.222</td>
<td class="has-text-centered">0.309</td>
<td class="has-text-centered">0.661</td>
<td class="has-text-centered">0.662</td>
<td class="has-text-centered">0.661</td>
<td class="has-text-centered">0.652</td>
<td class="has-text-centered">0.430</td>
<td class="has-text-centered">0.240</td>
<td class="has-text-centered performance-medium">0.980</td>
<td class="has-text-centered">0.386</td>
<td class="has-text-centered">0.811</td>
</tr>
<tr>
<td>Llama 3 8B Instruct</td>
<td class="has-text-centered">0.534</td>
<td class="has-text-centered">0.672</td>
<td class="has-text-centered">0.534</td>
<td class="has-text-centered">0.512</td>
<td class="has-text-centered">0.543</td>
<td class="has-text-centered">0.857</td>
<td class="has-text-centered">0.543</td>
<td class="has-text-centered">0.659</td>
<td class="has-text-centered">0.565</td>
<td class="has-text-centered">0.618</td>
<td class="has-text-centered">0.565</td>
<td class="has-text-centered">0.497</td>
<td class="has-text-centered">0.801</td>
<td class="has-text-centered">0.463</td>
<td class="has-text-centered">0.571</td>
<td class="has-text-centered">0.511</td>
<td class="has-text-centered">0.763</td>
</tr>
<tr>
<td>DBRX Instruct</td>
<td class="has-text-centered">0.578</td>
<td class="has-text-centered">0.706</td>
<td class="has-text-centered">0.578</td>
<td class="has-text-centered">0.574</td>
<td class="has-text-centered">0.359</td>
<td class="has-text-centered">0.851</td>
<td class="has-text-centered">0.359</td>
<td class="has-text-centered">0.483</td>
<td class="has-text-centered">0.285</td>
<td class="has-text-centered">0.572</td>
<td class="has-text-centered">0.285</td>
<td class="has-text-centered">0.193</td>
<td class="has-text-centered">0.222</td>
<td class="has-text-centered">0.190</td>
<td class="has-text-centered performance-best">1.000</td>
<td class="has-text-centered">0.319</td>
<td class="has-text-centered">0.746</td>
</tr>
<tr>
<td>DeepSeek LLM (67B)</td>
<td class="has-text-centered">0.596</td>
<td class="has-text-centered">0.711</td>
<td class="has-text-centered">0.596</td>
<td class="has-text-centered">0.578</td>
<td class="has-text-centered">0.369</td>
<td class="has-text-centered">0.856</td>
<td class="has-text-centered">0.369</td>
<td class="has-text-centered">0.492</td>
<td class="has-text-centered">0.532</td>
<td class="has-text-centered">0.678</td>
<td class="has-text-centered">0.532</td>
<td class="has-text-centered">0.407</td>
<td class="has-text-centered">0.832</td>
<td class="has-text-centered performance-best">1.000</td>
<td class="has-text-centered">0.082</td>
<td class="has-text-centered">0.151</td>
<td class="has-text-centered">0.778</td>
</tr>
<tr>
<td>Gemma 2 27B</td>
<td class="has-text-centered">0.639</td>
<td class="has-text-centered">0.730</td>
<td class="has-text-centered">0.639</td>
<td class="has-text-centered">0.621</td>
<td class="has-text-centered">0.410</td>
<td class="has-text-centered">0.849</td>
<td class="has-text-centered">0.410</td>
<td class="has-text-centered">0.538</td>
<td class="has-text-centered">0.651</td>
<td class="has-text-centered">0.704</td>
<td class="has-text-centered">0.651</td>
<td class="has-text-centered">0.620</td>
<td class="has-text-centered">0.471</td>
<td class="has-text-centered">0.257</td>
<td class="has-text-centered performance-best">1.000</td>
<td class="has-text-centered">0.408</td>
<td class="has-text-centered">0.808</td>
</tr>
<tr>
<td>Gemma 2 9B</td>
<td class="has-text-centered">0.630</td>
<td class="has-text-centered">0.710</td>
<td class="has-text-centered">0.630</td>
<td class="has-text-centered">0.609</td>
<td class="has-text-centered">0.412</td>
<td class="has-text-centered">0.848</td>
<td class="has-text-centered">0.412</td>
<td class="has-text-centered">0.541</td>
<td class="has-text-centered">0.595</td>
<td class="has-text-centered">0.694</td>
<td class="has-text-centered">0.595</td>
<td class="has-text-centered">0.519</td>
<td class="has-text-centered">0.371</td>
<td class="has-text-centered">0.224</td>
<td class="has-text-centered performance-strong">0.990</td>
<td class="has-text-centered">0.365</td>
<td class="has-text-centered performance-best">0.856</td>
</tr>
<tr>
<td>Mistral (7B) Instruct v0.3</td>
<td class="has-text-centered">0.547</td>
<td class="has-text-centered">0.677</td>
<td class="has-text-centered">0.547</td>
<td class="has-text-centered">0.528</td>
<td class="has-text-centered">0.375</td>
<td class="has-text-centered">0.839</td>
<td class="has-text-centered">0.375</td>
<td class="has-text-centered">0.503</td>
<td class="has-text-centered">0.587</td>
<td class="has-text-centered">0.598</td>
<td class="has-text-centered">0.587</td>
<td class="has-text-centered">0.542</td>
<td class="has-text-centered">0.521</td>
<td class="has-text-centered">0.266</td>
<td class="has-text-centered">0.918</td>
<td class="has-text-centered">0.412</td>
<td class="has-text-centered">0.779</td>
</tr>
<tr>
<td>Mixtral-8x22B Instruct</td>
<td class="has-text-centered">0.622</td>
<td class="has-text-centered">0.718</td>
<td class="has-text-centered">0.622</td>
<td class="has-text-centered">0.602</td>
<td class="has-text-centered">0.166</td>
<td class="has-text-centered">0.811</td>
<td class="has-text-centered">0.166</td>
<td class="has-text-centered">0.221</td>
<td class="has-text-centered">0.562</td>
<td class="has-text-centered">0.709</td>
<td class="has-text-centered">0.562</td>
<td class="has-text-centered">0.465</td>
<td class="has-text-centered">0.732</td>
<td class="has-text-centered">0.384</td>
<td class="has-text-centered">0.775</td>
<td class="has-text-centered">0.513</td>
<td class="has-text-centered performance-medium">0.835</td>
</tr>
<tr>
<td>Mixtral-8x7B Instruct</td>
<td class="has-text-centered">0.567</td>
<td class="has-text-centered">0.693</td>
<td class="has-text-centered">0.567</td>
<td class="has-text-centered">0.547</td>
<td class="has-text-centered">0.285</td>
<td class="has-text-centered">0.838</td>
<td class="has-text-centered">0.285</td>
<td class="has-text-centered">0.396</td>
<td class="has-text-centered">0.623</td>
<td class="has-text-centered">0.636</td>
<td class="has-text-centered">0.623</td>
<td class="has-text-centered">0.603</td>
<td class="has-text-centered">0.765</td>
<td class="has-text-centered">0.431</td>
<td class="has-text-centered">0.898</td>
<td class="has-text-centered">0.583</td>
<td class="has-text-centered">0.805</td>
</tr>
<tr>
<td>Qwen 2 Instruct (72B)</td>
<td class="has-text-centered">0.644</td>
<td class="has-text-centered">0.730</td>
<td class="has-text-centered">0.644</td>
<td class="has-text-centered">0.627</td>
<td class="has-text-centered">0.370</td>
<td class="has-text-centered">0.848</td>
<td class="has-text-centered">0.370</td>
<td class="has-text-centered">0.495</td>
<td class="has-text-centered">0.623</td>
<td class="has-text-centered">0.639</td>
<td class="has-text-centered">0.623</td>
<td class="has-text-centered">0.605</td>
<td class="has-text-centered">0.821</td>
<td class="has-text-centered">0.506</td>
<td class="has-text-centered">0.867</td>
<td class="has-text-centered">0.639</td>
<td class="has-text-centered">0.830</td>
</tr>
<tr>
<td>WizardLM-2 8x22B</td>
<td class="has-text-centered">0.664</td>
<td class="has-text-centered">0.737</td>
<td class="has-text-centered">0.664</td>
<td class="has-text-centered">0.648</td>
<td class="has-text-centered">0.373</td>
<td class="has-text-centered">0.842</td>
<td class="has-text-centered">0.373</td>
<td class="has-text-centered">0.500</td>
<td class="has-text-centered">0.583</td>
<td class="has-text-centered performance-medium">0.710</td>
<td class="has-text-centered">0.583</td>
<td class="has-text-centered">0.505</td>
<td class="has-text-centered">0.831</td>
<td class="has-text-centered">0.630</td>
<td class="has-text-centered">0.173</td>
<td class="has-text-centered">0.272</td>
<td class="has-text-centered">0.797</td>
</tr>
<tr>
<td>DeepSeek-V3</td>
<td class="has-text-centered performance-strong">0.722</td>
<td class="has-text-centered performance-medium">0.774</td>
<td class="has-text-centered performance-strong">0.722</td>
<td class="has-text-centered performance-strong">0.714</td>
<td class="has-text-centered">0.362</td>
<td class="has-text-centered">0.845</td>
<td class="has-text-centered">0.362</td>
<td class="has-text-centered">0.487</td>
<td class="has-text-centered">0.625</td>
<td class="has-text-centered performance-strong">0.712</td>
<td class="has-text-centered">0.625</td>
<td class="has-text-centered">0.578</td>
<td class="has-text-centered">0.860</td>
<td class="has-text-centered">0.586</td>
<td class="has-text-centered">0.796</td>
<td class="has-text-centered">0.675</td>
<td class="has-text-centered">0.729</td>
</tr>
<tr>
<td>DeepSeek R1</td>
<td class="has-text-centered performance-best">0.772</td>
<td class="has-text-centered performance-strong">0.789</td>
<td class="has-text-centered performance-best">0.772</td>
<td class="has-text-centered performance-best">0.763</td>
<td class="has-text-centered">0.306</td>
<td class="has-text-centered">0.846</td>
<td class="has-text-centered">0.306</td>
<td class="has-text-centered">0.419</td>
<td class="has-text-centered performance-strong">0.679</td>
<td class="has-text-centered">0.682</td>
<td class="has-text-centered performance-strong">0.679</td>
<td class="has-text-centered performance-strong">0.670</td>
<td class="has-text-centered">0.851</td>
<td class="has-text-centered">0.557</td>
<td class="has-text-centered">0.898</td>
<td class="has-text-centered">0.688</td>
<td class="has-text-centered">0.769</td>
</tr>
<tr>
<td>QwQ-32B-Preview</td>
<td class="has-text-centered">0.577</td>
<td class="has-text-centered">0.747</td>
<td class="has-text-centered">0.577</td>
<td class="has-text-centered">0.613</td>
<td class="has-text-centered performance-strong">0.716</td>
<td class="has-text-centered performance-strong">0.871</td>
<td class="has-text-centered performance-strong">0.716</td>
<td class="has-text-centered performance-strong">0.784</td>
<td class="has-text-centered">0.591</td>
<td class="has-text-centered">0.630</td>
<td class="has-text-centered">0.591</td>
<td class="has-text-centered">0.555</td>
<td class="has-text-centered">0.819</td>
<td class="has-text-centered performance-best">1.000</td>
<td class="has-text-centered">0.010</td>
<td class="has-text-centered">0.020</td>
<td class="has-text-centered">0.744</td>
</tr>
<tr>
<td>Jamba 1.5 Mini</td>
<td class="has-text-centered">0.528</td>
<td class="has-text-centered">0.630</td>
<td class="has-text-centered">0.528</td>
<td class="has-text-centered">0.508</td>
<td class="has-text-centered performance-best">0.913</td>
<td class="has-text-centered performance-best">0.883</td>
<td class="has-text-centered performance-best">0.913</td>
<td class="has-text-centered performance-best">0.898</td>
<td class="has-text-centered">0.572</td>
<td class="has-text-centered">0.678</td>
<td class="has-text-centered">0.572</td>
<td class="has-text-centered">0.499</td>
<td class="has-text-centered">0.812</td>
<td class="has-text-centered">0.429</td>
<td class="has-text-centered">0.092</td>
<td class="has-text-centered">0.151</td>
<td class="has-text-centered">0.682</td>
</tr>
<tr>
<td>Jamba 1.5 Large</td>
<td class="has-text-centered">0.642</td>
<td class="has-text-centered">0.746</td>
<td class="has-text-centered">0.642</td>
<td class="has-text-centered">0.628</td>
<td class="has-text-centered">0.494</td>
<td class="has-text-centered">0.851</td>
<td class="has-text-centered">0.494</td>
<td class="has-text-centered">0.618</td>
<td class="has-text-centered">0.597</td>
<td class="has-text-centered">0.650</td>
<td class="has-text-centered">0.597</td>
<td class="has-text-centered">0.550</td>
<td class="has-text-centered">0.855</td>
<td class="has-text-centered">0.639</td>
<td class="has-text-centered">0.469</td>
<td class="has-text-centered">0.541</td>
<td class="has-text-centered">0.782</td>
</tr>
<tr>
<td>Claude 3.5 Sonnet</td>
<td class="has-text-centered">0.682</td>
<td class="has-text-centered">0.755</td>
<td class="has-text-centered">0.682</td>
<td class="has-text-centered">0.668</td>
<td class="has-text-centered">0.513</td>
<td class="has-text-centered">0.854</td>
<td class="has-text-centered">0.513</td>
<td class="has-text-centered">0.634</td>
<td class="has-text-centered performance-medium">0.675</td>
<td class="has-text-centered">0.677</td>
<td class="has-text-centered performance-medium">0.675</td>
<td class="has-text-centered performance-best">0.674</td>
<td class="has-text-centered performance-medium">0.879</td>
<td class="has-text-centered">0.646</td>
<td class="has-text-centered">0.745</td>
<td class="has-text-centered performance-medium">0.692</td>
<td class="has-text-centered">0.827</td>
</tr>
<tr>
<td>Claude 3 Haiku</td>
<td class="has-text-centered">0.639</td>
<td class="has-text-centered">0.735</td>
<td class="has-text-centered">0.639</td>
<td class="has-text-centered">0.622</td>
<td class="has-text-centered">0.067</td>
<td class="has-text-centered">0.674</td>
<td class="has-text-centered">0.067</td>
<td class="has-text-centered">0.022</td>
<td class="has-text-centered">0.633</td>
<td class="has-text-centered">0.634</td>
<td class="has-text-centered">0.633</td>
<td class="has-text-centered">0.631</td>
<td class="has-text-centered">0.838</td>
<td class="has-text-centered">0.556</td>
<td class="has-text-centered">0.561</td>
<td class="has-text-centered">0.558</td>
<td class="has-text-centered">0.781</td>
</tr>
<tr>
<td>Cohere Command R 7B</td>
<td class="has-text-centered">0.530</td>
<td class="has-text-centered">0.650</td>
<td class="has-text-centered">0.530</td>
<td class="has-text-centered">0.516</td>
<td class="has-text-centered performance-medium">0.682</td>
<td class="has-text-centered performance-medium">0.868</td>
<td class="has-text-centered performance-medium">0.682</td>
<td class="has-text-centered performance-medium">0.762</td>
<td class="has-text-centered">0.536</td>
<td class="has-text-centered">0.505</td>
<td class="has-text-centered">0.536</td>
<td class="has-text-centered">0.459</td>
<td class="has-text-centered">0.797</td>
<td class="has-text-centered">0.210</td>
<td class="has-text-centered">0.041</td>
<td class="has-text-centered">0.068</td>
<td class="has-text-centered">0.770</td>
</tr>
<tr>
<td>Cohere Command R +</td>
<td class="has-text-centered">0.660</td>
<td class="has-text-centered">0.747</td>
<td class="has-text-centered">0.660</td>
<td class="has-text-centered">0.651</td>
<td class="has-text-centered">0.575</td>
<td class="has-text-centered">0.859</td>
<td class="has-text-centered">0.575</td>
<td class="has-text-centered">0.684</td>
<td class="has-text-centered">0.526</td>
<td class="has-text-centered">0.655</td>
<td class="has-text-centered">0.526</td>
<td class="has-text-centered">0.393</td>
<td class="has-text-centered">0.804</td>
<td class="has-text-centered">0.333</td>
<td class="has-text-centered">0.071</td>
<td class="has-text-centered">0.118</td>
<td class="has-text-centered">0.812</td>
</tr>
<tr>
<td>Google Gemini 1.5 Pro</td>
<td class="has-text-centered">0.483</td>
<td class="has-text-centered">0.487</td>
<td class="has-text-centered">0.483</td>
<td class="has-text-centered">0.418</td>
<td class="has-text-centered">0.240</td>
<td class="has-text-centered">0.823</td>
<td class="has-text-centered">0.240</td>
<td class="has-text-centered">0.336</td>
<td class="has-text-centered">0.619</td>
<td class="has-text-centered">0.667</td>
<td class="has-text-centered">0.619</td>
<td class="has-text-centered">0.579</td>
<td class="has-text-centered">0.700</td>
<td class="has-text-centered">0.369</td>
<td class="has-text-centered">0.908</td>
<td class="has-text-centered">0.525</td>
<td class="has-text-centered performance-strong">0.837</td>
</tr>
<tr>
<td>OpenAI gpt-4o</td>
<td class="has-text-centered performance-medium">0.704</td>
<td class="has-text-centered performance-best">0.792</td>
<td class="has-text-centered performance-medium">0.704</td>
<td class="has-text-centered performance-medium">0.710</td>
<td class="has-text-centered">0.396</td>
<td class="has-text-centered">0.846</td>
<td class="has-text-centered">0.396</td>
<td class="has-text-centered">0.524</td>
<td class="has-text-centered performance-best">0.681</td>
<td class="has-text-centered performance-best">0.719</td>
<td class="has-text-centered performance-best">0.681</td>
<td class="has-text-centered performance-medium">0.664</td>
<td class="has-text-centered performance-best">0.896</td>
<td class="has-text-centered performance-medium">0.667</td>
<td class="has-text-centered">0.857</td>
<td class="has-text-centered performance-best">0.750</td>
<td class="has-text-centered">0.824</td>
</tr>
<tr>
<td>OpenAI o1-mini</td>
<td class="has-text-centered">0.681</td>
<td class="has-text-centered">0.760</td>
<td class="has-text-centered">0.681</td>
<td class="has-text-centered">0.670</td>
<td class="has-text-centered">0.487</td>
<td class="has-text-centered">0.851</td>
<td class="has-text-centered">0.487</td>
<td class="has-text-centered">0.612</td>
<td class="has-text-centered">0.651</td>
<td class="has-text-centered">0.670</td>
<td class="has-text-centered">0.651</td>
<td class="has-text-centered">0.635</td>
<td class="has-text-centered performance-strong">0.888</td>
<td class="has-text-centered performance-medium">0.664</td>
<td class="has-text-centered">0.786</td>
<td class="has-text-centered performance-strong">0.720</td>
<td class="has-text-centered">0.769</td>
</tr>
</tbody>
</table>
<div class="content is-small mt-4">
<p><strong>Note:</strong> Color highlighting indicates performance ranking:
<span class="performance-best"> Best </span>,
<span class="performance-medium"> Strong </span>,
<span class="performance-low"> Good </span>
</p>
</div>
</div>
</div>
<!-- Text Summarization tab content -->
<div id="text-summarization" class="tab-content">
<h2 class="title is-4">Text Summarization Task Results</h2>
<div class="results-table">
<table class="table is-bordered is-striped is-narrow is-hoverable is-fullwidth">
<thead>
<tr>
<th rowspan="2">Model</th>
<th colspan="3" class="has-text-centered tooltip-trigger tooltip-right" data-title="ECTSum" data-tooltip="ECTSum contains 2,425 document-summary pairs featuring earnings call transcripts paired with concise bullet-point summaries extracted from Reuters articles. The summarization task requires extracting and condensing key financial information from lengthy corporate communications into brief, informative points.">ECTSum</th>
<th colspan="3" class="has-text-centered tooltip-trigger tooltip-right" data-title="EDTSum" data-tooltip="EDTSum consists of 2,000 financial news articles paired with their headlines as ground-truth summaries for evaluating text summarization. The task challenges models to condense complex financial news articles into concise, informative headlines that capture the essential information.">EDTSum</th>
</tr>
<tr>
<th class="has-text-centered">BERTScore Precision</th>
<th class="has-text-centered">BERTScore Recall</th>
<th class="has-text-centered">BERTScore F1</th>
<th class="has-text-centered">BERTScore Precision</th>
<th class="has-text-centered">BERTScore Recall</th>
<th class="has-text-centered">BERTScore F1</th>
</tr>
</thead>
<tbody>
<tr>
<td>Llama 3 70B Instruct</td>
<td class="has-text-centered">0.715</td>
<td class="has-text-centered">0.801</td>
<td class="has-text-centered">0.754</td>
<td class="has-text-centered">0.793</td>
<td class="has-text-centered performance-medium">0.844</td>
<td class="has-text-centered performance-strong">0.817</td>
</tr>
<tr>
<td>Llama 3 8B Instruct</td>
<td class="has-text-centered">0.724</td>
<td class="has-text-centered">0.796</td>
<td class="has-text-centered">0.757</td>
<td class="has-text-centered">0.785</td>
<td class="has-text-centered">0.841</td>
<td class="has-text-centered">0.811</td>
</tr>
<tr>
<td>DBRX Instruct</td>
<td class="has-text-centered">0.680</td>
<td class="has-text-centered">0.786</td>
<td class="has-text-centered">0.729</td>
<td class="has-text-centered">0.774</td>
<td class="has-text-centered">0.843</td>
<td class="has-text-centered">0.806</td>
</tr>
<tr>
<td>DeepSeek LLM (67B)</td>
<td class="has-text-centered">0.692</td>
<td class="has-text-centered">0.678</td>
<td class="has-text-centered">0.681</td>
<td class="has-text-centered">0.779</td>
<td class="has-text-centered">0.840</td>
<td class="has-text-centered">0.807</td>
</tr>
<tr>
<td>Gemma 2 27B</td>
<td class="has-text-centered">0.680</td>
<td class="has-text-centered">0.777</td>
<td class="has-text-centered">0.723</td>
<td class="has-text-centered performance-strong">0.801</td>
<td class="has-text-centered">0.829</td>
<td class="has-text-centered">0.814</td>
</tr>
<tr>
<td>Gemma 2 9B</td>
<td class="has-text-centered">0.651</td>
<td class="has-text-centered">0.531</td>
<td class="has-text-centered">0.585</td>
<td class="has-text-centered performance-best">0.803</td>
<td class="has-text-centered">0.833</td>
<td class="has-text-centered performance-strong">0.817</td>
</tr>
<tr>
<td>Mistral (7B) Instruct v0.3</td>
<td class="has-text-centered">0.702</td>
<td class="has-text-centered performance-strong">0.806</td>
<td class="has-text-centered">0.750</td>
<td class="has-text-centered">0.783</td>
<td class="has-text-centered">0.842</td>
<td class="has-text-centered">0.811</td>
</tr>
<tr>
<td>Mixtral-8x22B Instruct</td>
<td class="has-text-centered">0.713</td>
<td class="has-text-centered performance-best">0.812</td>
<td class="has-text-centered">0.758</td>
<td class="has-text-centered">0.790</td>
<td class="has-text-centered">0.843</td>
<td class="has-text-centered">0.815</td>
</tr>
<tr>
<td>Mixtral-8x7B Instruct</td>
<td class="has-text-centered">0.727</td>
<td class="has-text-centered">0.773</td>
<td class="has-text-centered">0.747</td>
<td class="has-text-centered">0.785</td>
<td class="has-text-centered">0.839</td>
<td class="has-text-centered">0.810</td>
</tr>
<tr>
<td>Qwen 2 Instruct (72B)</td>
<td class="has-text-centered">0.709</td>
<td class="has-text-centered performance-medium">0.804</td>
<td class="has-text-centered">0.752</td>
<td class="has-text-centered">0.781</td>
<td class="has-text-centered performance-strong">0.846</td>
<td class="has-text-centered">0.811</td>
</tr>
<tr>
<td>WizardLM-2 8x22B</td>
<td class="has-text-centered">0.677</td>
<td class="has-text-centered performance-strong">0.806</td>
<td class="has-text-centered">0.735</td>
<td class="has-text-centered">0.774</td>
<td class="has-text-centered performance-best">0.847</td>
<td class="has-text-centered">0.808</td>
</tr>
<tr>
<td>DeepSeek-V3</td>
<td class="has-text-centered">0.703</td>
<td class="has-text-centered performance-strong">0.806</td>
<td class="has-text-centered">0.750</td>
<td class="has-text-centered">0.791</td>
<td class="has-text-centered">0.842</td>
<td class="has-text-centered">0.815</td>
</tr>
<tr>
<td>DeepSeek R1</td>
<td class="has-text-centered">0.724</td>
<td class="has-text-centered">0.800</td>
<td class="has-text-centered">0.759</td>
<td class="has-text-centered">0.770</td>
<td class="has-text-centered">0.843</td>
<td class="has-text-centered">0.804</td>
</tr>
<tr>
<td>QwQ-32B-Preview</td>
<td class="has-text-centered">0.653</td>
<td class="has-text-centered">0.751</td>
<td class="has-text-centered">0.696</td>
<td class="has-text-centered">0.797</td>
<td class="has-text-centered">0.841</td>
<td class="has-text-centered performance-strong">0.817</td>
</tr>
<tr>
<td>Jamba 1.5 Mini</td>
<td class="has-text-centered">0.692</td>
<td class="has-text-centered">0.798</td>
<td class="has-text-centered">0.741</td>
<td class="has-text-centered">0.798</td>
<td class="has-text-centered">0.838</td>
<td class="has-text-centered performance-medium">0.816</td>
</tr>
<tr>
<td>Jamba 1.5 Large</td>
<td class="has-text-centered">0.679</td>
<td class="has-text-centered">0.800</td>
<td class="has-text-centered">0.734</td>
<td class="has-text-centered">0.799</td>
<td class="has-text-centered">0.841</td>
<td class="has-text-centered performance-best">0.818</td>
</tr>
<tr>
<td>Claude 3.5 Sonnet</td>
<td class="has-text-centered performance-medium">0.737</td>
<td class="has-text-centered">0.802</td>
<td class="has-text-centered performance-medium">0.767</td>
<td class="has-text-centered">0.786</td>
<td class="has-text-centered">0.843</td>
<td class="has-text-centered">0.813</td>
</tr>
<tr>
<td>Claude 3 Haiku</td>
<td class="has-text-centered">0.683</td>
<td class="has-text-centered">0.617</td>
<td class="has-text-centered">0.646</td>
<td class="has-text-centered">0.778</td>
<td class="has-text-centered performance-medium">0.844</td>
<td class="has-text-centered">0.808</td>
</tr>
<tr>
<td>Cohere Command R 7B</td>
<td class="has-text-centered">0.724</td>
<td class="has-text-centered">0.781</td>
<td class="has-text-centered">0.750</td>
<td class="has-text-centered">0.790</td>
<td class="has-text-centered performance-medium">0.844</td>
<td class="has-text-centered">0.815</td>
</tr>
<tr>
<td>Cohere Command R +</td>
<td class="has-text-centered">0.724</td>
<td class="has-text-centered">0.782</td>
<td class="has-text-centered">0.751</td>
<td class="has-text-centered">0.789</td>
<td class="has-text-centered">0.834</td>
<td class="has-text-centered">0.810</td>
</tr>
<tr>
<td>Google Gemini 1.5 Pro</td>
<td class="has-text-centered performance-best">0.757</td>
<td class="has-text-centered">0.800</td>
<td class="has-text-centered performance-best">0.777</td>
<td class="has-text-centered performance-medium">0.800</td>
<td class="has-text-centered">0.836</td>
<td class="has-text-centered performance-strong">0.817</td>
</tr>
<tr>
<td>OpenAI gpt-4o</td>
<td class="has-text-centered performance-strong">0.755</td>
<td class="has-text-centered">0.793</td>
<td class="has-text-centered performance-strong">0.773</td>
<td class="has-text-centered">0.795</td>
<td class="has-text-centered">0.840</td>
<td class="has-text-centered performance-medium">0.816</td>
</tr>
<tr>
<td>OpenAI o1-mini</td>
<td class="has-text-centered">0.731</td>
<td class="has-text-centered">0.801</td>
<td class="has-text-centered">0.763</td>
<td class="has-text-centered">0.795</td>
<td class="has-text-centered">0.840</td>
<td class="has-text-centered performance-medium">0.816</td>
</tr>
</tbody>
</table>
<div class="content is-small mt-4">
<p><strong>Note:</strong> Color highlighting indicates performance ranking:
<span class="performance-best"> Best </span>,
<span class="performance-medium"> Strong </span>,
<span class="performance-low"> Good </span>
</p>
</div>
</div>
</div>
<!-- Cost Analysis Tab -->
<div id="cost-analysis" class="tab-content">
<h2 class="title is-4">Model Cost Analysis</h2>
<div class="results-table">
<table class="table is-bordered is-striped is-narrow is-hoverable is-fullwidth">
<thead>
<tr>
<th>Model</th>
<th class="has-text-centered tooltip-trigger column-border-left" data-title="FOMC" data-tooltip="FOMC is a dataset containing Federal Open Market Committee speeches, meeting minutes, and press conference transcripts spanning from 1996 to 2022. The classification task involves determining whether the monetary policy stance expressed in each document is hawkish (tighter monetary policy) or dovish (looser monetary policy).">FOMC</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="Financial Phrase Bank" data-tooltip="Financial Phrase Bank (FPB) contains 4,840 sentences from financial news articles categorized as positive, negative, or neutral by 16 finance experts using majority voting. The sentiment classification task requires understanding how these statements might influence investor perception of stock prices.">FPB</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="FinQA" data-tooltip="FinQA contains 8,281 question-answer pairs derived from financial reports that require numerical reasoning over tabular financial data. The question-answering task features multi-step reasoning challenges with full annotation of reasoning programs to solve complex financial queries.">FinQA</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="FiQA Task 1" data-tooltip="FiQA Task 1 focuses on aspect-based financial sentiment analysis in microblog posts and news headlines using a continuous scale from -1 (negative) to 1 (positive). The regression task requires models to accurately predict the sentiment score that reflects investor perception of financial texts.">FiQA-1</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="FiQA Task 2" data-tooltip="FiQA Task 2 focuses on sentiment analysis of financial texts from various sources.">FiQA-2</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="Headlines" data-tooltip="Headlines is a dataset containing 11,412 human-annotated financial news headlines focused on commodities, particularly gold, spanning from 2000 to 2019. The classification task involves identifying binary indicators for price mentions and directional price movements in these concise financial texts.">HL</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="FinBench" data-tooltip="FinBench is a comprehensive evaluation dataset containing 333,000 labeled instances that combines tabular data and profile text for financial risk prediction. The task requires models to predict financial outcomes across three key risk categories: default, fraud, and customer churn.">FB</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="FinRED" data-tooltip="FinRED is a specialized relation extraction dataset created from financial news and earnings call transcripts using distance supervision based on Wikidata triplets. The task involves identifying and extracting financial relationships between entities to understand connections in financial contexts.">FR</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="ReFiND" data-tooltip="ReFiND is a comprehensive relation extraction dataset containing approximately 29,000 annotated instances with 22 distinct relation types across 8 entity pair categories from various financial documents. The task requires identifying specific relationships between financial entities in complex documents like SEC filings.">RD</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="EDTSum" data-tooltip="EDTSum consists of 2,000 financial news articles paired with their headlines as ground-truth summaries for evaluating text summarization. The task challenges models to condense complex financial news articles into concise, informative headlines that capture the essential information.">EDTSum</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="Banking77" data-tooltip="Banking77 is a fine-grained dataset comprising 13,083 customer service queries annotated with 77 unique intents from the banking domain. The task involves accurately classifying each customer query into the correct intent category to improve automated banking support systems.">B77</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="FinCausal (CD)" data-tooltip="FinCausal Causal Discovery (CD) contains 29,444 text sections from financial news, with 2,136 annotated as expressing causal relationships. The task involves extracting precise cause and effect spans from financial texts that contain causal relationships.">CD</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="FinCausal (CC)" data-tooltip="FinCausal Causality Classification (CC) consists of 29,444 text sections from financial news with binary annotations indicating causal relationships. The classification task requires determining whether a given financial text section contains a causal relationship (1) or not (0).">CC</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="ECTSum" data-tooltip="ECTSum contains 2,425 document-summary pairs featuring earnings call transcripts paired with concise bullet-point summaries extracted from Reuters articles. The summarization task requires extracting and condensing key financial information from lengthy corporate communications into brief, informative points.">ECTSum</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="FinEntity" data-tooltip="FinEntity consists of 979 financial news paragraphs containing 2,131 manually-annotated financial entities with sentiment classifications. The task involves identifying companies and asset classes in financial texts while determining the associated sentiment expressed toward each entity.">FE</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="FiNER-ORD" data-tooltip="FiNER-ORD is a manually annotated named entity recognition dataset comprising financial news articles with detailed entity annotations. The task requires identifying and correctly classifying person, location, and organization entities in financial contexts.">FiNER</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="FNXL" data-tooltip="FNXL contains 79,088 sentences with 142,922 annotated numerals extracted from SEC 10-K reports and categorized under 2,794 distinct numerical labels. The information extraction task requires identifying, categorizing and understanding the financial significance of numerical entities in regulatory filings.">FNXL</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="NumClaim" data-tooltip="NumClaim is an expert-annotated dataset for detecting and analyzing fine-grained investor claims within financial narratives that contain numerical information. The task requires identifying and categorizing claims containing numerals in analyst reports and earnings call transcripts for investment decision making.">NC</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="TATQA" data-tooltip="TATQA is a large-scale question answering dataset for hybrid data sources that combines tables and text from financial reports. The task emphasizes numerical reasoning operations across multiple formats, requiring models to integrate information from structured and unstructured sources to answer financial questions.">TQA</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="ConvFinQA" data-tooltip="ConvFinQA is a multi-turn question answering dataset with 3,892 conversations containing 14,115 questions that explore chains of numerical reasoning in financial contexts. The conversational task requires maintaining context while performing sequential numerical operations to answer increasingly complex financial questions.">CFQA</th>
<th class="has-text-centered tooltip-trigger tooltip-right" data-title="SubjECTive-QA" data-tooltip="SubjECTive-QA contains 49,446 annotations across 2,747 question-answer pairs extracted from 120 earnings call transcripts. The multi-label classification task involves analyzing six subjective features in financial discourse: assertiveness, cautiousness, optimism, specificity, clarity, and relevance.">SQA</th>
<th class="has-text-centered tooltip-trigger tooltip-right column-border-left" data-title="Total Cost" data-tooltip="Total cost in USD to run inference on all datasets combined">Total</th>
</tr>
</thead>
<tbody>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="Llama 3 70B Instruct" data-tooltip="Meta's advanced 70 billion parameter dense language model optimized for instruction-following tasks. Available through Together AI and notable for complex reasoning capabilities.">Llama 3 70B Instruct</td>
<td class="column-border-left has-text-centered">0.10</td>
<td class="has-text-centered">0.11</td>
<td class="has-text-centered">1.14</td>
<td class="has-text-centered">0.06</td>
<td class="has-text-centered">0.72</td>
<td class="has-text-centered">1.00</td>
<td class="has-text-centered">0.40</td>
<td class="has-text-centered">0.38</td>
<td class="has-text-centered">1.34</td>
<td class="has-text-centered">1.94</td>
<td class="has-text-centered">1.64</td>
<td class="has-text-centered">0.07</td>
<td class="has-text-centered">0.05</td>
<td class="has-text-centered">1.56</td>
<td class="has-text-centered">0.12</td>
<td class="has-text-centered">0.33</td>
<td class="has-text-centered">0.25</td>
<td class="has-text-centered">0.09</td>
<td class="has-text-centered">1.11</td>
<td class="has-text-centered">2.96</td>
<td class="has-text-centered">1.17</td>
<td class="column-border-left has-text-centered performance-medium">16.54</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="Llama 3 8B Instruct" data-tooltip="Meta's compact 8 billion parameter language model, offering an efficient alternative to larger variants. Optimized for practical instruction-following tasks with reduced computational requirements.">Llama 3 8B Instruct</td>
<td class="column-border-left has-text-centered">0.02</td>
<td class="has-text-centered">0.03</td>
<td class="has-text-centered">0.25</td>
<td class="has-text-centered">0.01</td>
<td class="has-text-centered">0.16</td>
<td class="has-text-centered">0.22</td>
<td class="has-text-centered">0.09</td>
<td class="has-text-centered">0.09</td>
<td class="has-text-centered">0.32</td>
<td class="has-text-centered">0.43</td>
<td class="has-text-centered">0.37</td>
<td class="has-text-centered">0.02</td>
<td class="has-text-centered">0.01</td>
<td class="has-text-centered">0.36</td>
<td class="has-text-centered">0.03</td>
<td class="has-text-centered">0.08</td>
<td class="has-text-centered">0.06</td>
<td class="has-text-centered">0.02</td>
<td class="has-text-centered">0.26</td>
<td class="has-text-centered">0.69</td>
<td class="has-text-centered">0.26</td>
<td class="column-border-left has-text-centered performance-low">3.79</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="DBRX Instruct" data-tooltip="A 132 billion parameter Mixture of Experts (MoE) model developed by Databricks. Fine-tuned for instruction-following on the Mosaic AI platform with strong performance across diverse tasks.">DBRX Instruct</td>
<td class="column-border-left has-text-centered">0.14</td>
<td class="has-text-centered">0.17</td>
<td class="has-text-centered">1.50</td>
<td class="has-text-centered">0.06</td>
<td class="has-text-centered">0.95</td>
<td class="has-text-centered">1.29</td>
<td class="has-text-centered">0.56</td>
<td class="has-text-centered">0.57</td>
<td class="has-text-centered">2.05</td>
<td class="has-text-centered">2.93</td>
<td class="has-text-centered">2.14</td>
<td class="has-text-centered">0.11</td>
<td class="has-text-centered">0.10</td>
<td class="has-text-centered">2.45</td>
<td class="has-text-centered">0.17</td>
<td class="has-text-centered">0.47</td>
<td class="has-text-centered">0.34</td>
<td class="has-text-centered">0.13</td>
<td class="has-text-centered">1.47</td>
<td class="has-text-centered">4.19</td>
<td class="has-text-centered">1.55</td>
<td class="column-border-left has-text-centered performance-medium">23.35</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="DeepSeek LLM (67B)" data-tooltip="DeepSeek's 67 billion parameter dense language model specialized for complex reasoning and instruction following. Developed by the DeepSeek AI team to compete with other frontier models.">DeepSeek LLM (67B)</td>
<td class="column-border-left has-text-centered">0.10</td>
<td class="has-text-centered">0.12</td>
<td class="has-text-centered">1.25</td>
<td class="has-text-centered">0.05</td>
<td class="has-text-centered">0.76</td>
<td class="has-text-centered">0.87</td>
<td class="has-text-centered">0.42</td>
<td class="has-text-centered">0.37</td>
<td class="has-text-centered">1.45</td>
<td class="has-text-centered">1.85</td>
<td class="has-text-centered">2.03</td>
<td class="has-text-centered">0.08</td>
<td class="has-text-centered">0.05</td>
<td class="has-text-centered">0.83</td>
<td class="has-text-centered">0.13</td>
<td class="has-text-centered">0.34</td>
<td class="has-text-centered">0.24</td>
<td class="has-text-centered">0.09</td>
<td class="has-text-centered">1.20</td>
<td class="has-text-centered">3.17</td>
<td class="has-text-centered">1.17</td>
<td class="column-border-left has-text-centered performance-medium">16.57</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="Gemma 2 27B" data-tooltip="Google's 27 billion parameter language model released as an open-weight alternative to proprietary models. Designed for general NLP applications with strong performance-to-size ratio.">Gemma 2 27B</td>
<td class="column-border-left has-text-centered">0.08</td>
<td class="has-text-centered">0.09</td>
<td class="has-text-centered">1.05</td>
<td class="has-text-centered">0.05</td>
<td class="has-text-centered">0.66</td>
<td class="has-text-centered">0.91</td>
<td class="has-text-centered">0.30</td>
<td class="has-text-centered">0.34</td>
<td class="has-text-centered">1.37</td>
<td class="has-text-centered">1.75</td>
<td class="has-text-centered">1.77</td>
<td class="has-text-centered">0.07</td>
<td class="has-text-centered">0.04</td>
<td class="has-text-centered">1.46</td>
<td class="has-text-centered">0.11</td>
<td class="has-text-centered">0.30</td>
<td class="has-text-centered">0.21</td>
<td class="has-text-centered">0.08</td>
<td class="has-text-centered">1.00</td>
<td class="has-text-centered">2.84</td>
<td class="has-text-centered">1.04</td>
<td class="column-border-left has-text-centered performance-medium">15.50</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="Gemma 2 9B" data-tooltip="Google's compact 9 billion parameter language model offering an efficient alternative to larger models. Features strong performance-to-size ratio for general NLP tasks with reduced computing requirements.">Gemma 2 9B</td>
<td class="column-border-left has-text-centered">0.03</td>
<td class="has-text-centered">0.03</td>
<td class="has-text-centered">0.40</td>
<td class="has-text-centered">0.02</td>
<td class="has-text-centered">0.24</td>
<td class="has-text-centered">0.33</td>
<td class="has-text-centered">0.12</td>
<td class="has-text-centered">0.14</td>
<td class="has-text-centered">0.51</td>
<td class="has-text-centered">0.66</td>
<td class="has-text-centered">0.66</td>
<td class="has-text-centered">0.03</td>
<td class="has-text-centered">0.02</td>
<td class="has-text-centered">0.00</td>
<td class="has-text-centered">0.04</td>
<td class="has-text-centered">0.11</td>
<td class="has-text-centered">0.08</td>
<td class="has-text-centered">0.03</td>
<td class="has-text-centered">0.37</td>
<td class="has-text-centered">1.08</td>
<td class="has-text-centered">0.39</td>
<td class="column-border-left has-text-centered performance-low">5.29</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="Mistral (7B) Instruct v0.3" data-tooltip="Mistral AI's 7 billion parameter dense language model designed for instruction following. Notable for its efficiency and strong performance at a smaller size, making it suitable for cost-sensitive applications.">Mistral (7B) Instruct v0.3</td>
<td class="column-border-left has-text-centered">0.03</td>
<td class="has-text-centered">0.03</td>
<td class="has-text-centered">0.28</td>
<td class="has-text-centered">0.01</td>
<td class="has-text-centered">0.18</td>
<td class="has-text-centered">0.24</td>
<td class="has-text-centered">0.10</td>
<td class="has-text-centered">0.09</td>
<td class="has-text-centered">0.36</td>
<td class="has-text-centered">0.57</td>
<td class="has-text-centered">0.48</td>
<td class="has-text-centered">0.02</td>
<td class="has-text-centered">0.01</td>
<td class="has-text-centered">0.45</td>
<td class="has-text-centered">0.03</td>
<td class="has-text-centered">0.08</td>
<td class="has-text-centered">0.06</td>
<td class="has-text-centered">0.02</td>
<td class="has-text-centered">0.27</td>
<td class="has-text-centered">0.78</td>
<td class="has-text-centered">0.26</td>
<td class="column-border-left has-text-centered performance-low">4.36</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="Mixtral-8x22B Instruct" data-tooltip="Mistral AI's 141 billion parameter Mixture of Experts (MoE) model that combines eight 22B expert networks. Advanced architecture enables state-of-the-art performance while maintaining inference efficiency.">Mixtral-8x22B Instruct</td>
<td class="column-border-left has-text-centered">0.14</td>
<td class="has-text-centered">0.17</td>
<td class="has-text-centered">1.80</td>
<td class="has-text-centered">0.07</td>
<td class="has-text-centered">1.05</td>
<td class="has-text-centered">1.44</td>
<td class="has-text-centered">0.58</td>
<td class="has-text-centered">0.56</td>
<td class="has-text-centered">2.04</td>
<td class="has-text-centered">3.42</td>
<td class="has-text-centered">2.89</td>
<td class="has-text-centered">0.11</td>
<td class="has-text-centered">0.07</td>
<td class="has-text-centered">2.66</td>
<td class="has-text-centered">0.18</td>
<td class="has-text-centered">0.48</td>
<td class="has-text-centered">0.35</td>
<td class="has-text-centered">0.14</td>
<td class="has-text-centered">1.73</td>
<td class="has-text-centered">4.90</td>
<td class="has-text-centered">1.55</td>
<td class="column-border-left has-text-centered performance-medium">26.35</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="Mixtral-8x7B Instruct" data-tooltip="Mistral AI's 46.7 billion parameter Mixture of Experts (MoE) model combining eight 7B expert networks. Provides a balanced approach between performance and efficiency for instruction-following tasks.">Mixtral-8x7B Instruct</td>
<td class="column-border-left has-text-centered">0.08</td>
<td class="has-text-centered">0.09</td>
<td class="has-text-centered">0.88</td>
<td class="has-text-centered">0.04</td>
<td class="has-text-centered">0.53</td>
<td class="has-text-centered">0.70</td>
<td class="has-text-centered">0.30</td>
<td class="has-text-centered">0.30</td>
<td class="has-text-centered">1.07</td>
<td class="has-text-centered">1.72</td>
<td class="has-text-centered">1.50</td>
<td class="has-text-centered">0.06</td>
<td class="has-text-centered">0.05</td>
<td class="has-text-centered">1.30</td>
<td class="has-text-centered">0.09</td>
<td class="has-text-centered">0.24</td>
<td class="has-text-centered">0.20</td>
<td class="has-text-centered">0.07</td>
<td class="has-text-centered">0.87</td>
<td class="has-text-centered">2.55</td>
<td class="has-text-centered">0.78</td>
<td class="column-border-left has-text-centered performance-medium">13.41</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="Qwen 2 Instruct (72B)" data-tooltip="Alibaba Cloud's 72 billion parameter dense language model designed for instruction following. Features strong multilingual capabilities and optimization for diverse NLP tasks.">Qwen 2 Instruct (72B)</td>
<td class="column-border-left has-text-centered">0.10</td>
<td class="has-text-centered">0.12</td>
<td class="has-text-centered">1.29</td>
<td class="has-text-centered">0.05</td>
<td class="has-text-centered">0.74</td>
<td class="has-text-centered">0.96</td>
<td class="has-text-centered">0.43</td>
<td class="has-text-centered">0.43</td>
<td class="has-text-centered">1.44</td>
<td class="has-text-centered">2.36</td>
<td class="has-text-centered">1.61</td>
<td class="has-text-centered">0.08</td>
<td class="has-text-centered">0.05</td>
<td class="has-text-centered">1.80</td>
<td class="has-text-centered">0.12</td>
<td class="has-text-centered">0.34</td>
<td class="has-text-centered">0.24</td>
<td class="has-text-centered">0.10</td>
<td class="has-text-centered">1.18</td>
<td class="has-text-centered">3.41</td>
<td class="has-text-centered">1.17</td>
<td class="column-border-left has-text-centered performance-medium">18.02</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="WizardLM-2 8x22B" data-tooltip="Microsoft's 141 billion parameter Mixture of Experts (MoE) model specifically optimized for instruction-following tasks. Combines eight 22B expert networks with advanced routing for high performance.">WizardLM-2 8x22B</td>
<td class="column-border-left has-text-centered">0.16</td>
<td class="has-text-centered">0.19</td>
<td class="has-text-centered">1.94</td>
<td class="has-text-centered">0.08</td>
<td class="has-text-centered">1.07</td>
<td class="has-text-centered">1.47</td>
<td class="has-text-centered">0.61</td>
<td class="has-text-centered">0.61</td>
<td class="has-text-centered">2.24</td>
<td class="has-text-centered">3.47</td>
<td class="has-text-centered">3.00</td>
<td class="has-text-centered">0.11</td>
<td class="has-text-centered">0.10</td>
<td class="has-text-centered">2.85</td>
<td class="has-text-centered">0.18</td>
<td class="has-text-centered">0.49</td>
<td class="has-text-centered">0.34</td>
<td class="has-text-centered">0.14</td>
<td class="has-text-centered">1.94</td>
<td class="has-text-centered">5.31</td>
<td class="has-text-centered">1.55</td>
<td class="column-border-left has-text-centered performance-medium">27.87</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="DeepSeek-V3" data-tooltip="DeepSeek's 685 billion parameter Mixture of Experts (MoE) language model designed for advanced reasoning and instruction following. Features strong performance across diverse tasks with efficient architecture.">DeepSeek-V3</td>
<td class="column-border-left has-text-centered">0.13</td>
<td class="has-text-centered">0.15</td>
<td class="has-text-centered">1.57</td>
<td class="has-text-centered">0.07</td>
<td class="has-text-centered">0.98</td>
<td class="has-text-centered">1.36</td>
<td class="has-text-centered">0.52</td>
<td class="has-text-centered">0.54</td>
<td class="has-text-centered">2.10</td>
<td class="has-text-centered">2.99</td>
<td class="has-text-centered">2.55</td>
<td class="has-text-centered">0.11</td>
<td class="has-text-centered">0.06</td>
<td class="has-text-centered">2.33</td>
<td class="has-text-centered">0.16</td>
<td class="has-text-centered">0.55</td>
<td class="has-text-centered">0.28</td>
<td class="has-text-centered">0.12</td>
<td class="has-text-centered">1.56</td>
<td class="has-text-centered">4.28</td>
<td class="has-text-centered">1.62</td>
<td class="column-border-left has-text-centered performance-medium">24.03</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="DeepSeek R1" data-tooltip="DeepSeek's premium 671 billion parameter Mixture of Experts (MoE) model representing their most advanced offering. Designed for state-of-the-art performance across complex reasoning and financial tasks.">DeepSeek R1</td>
<td class="column-border-left has-text-centered">1.99</td>
<td class="has-text-centered">2.10</td>
<td class="has-text-centered">14.18</td>
<td class="has-text-centered">1.48</td>
<td class="has-text-centered">17.82</td>
<td class="has-text-centered">20.11</td>
<td class="has-text-centered">6.63</td>
<td class="has-text-centered">12.65</td>
<td class="has-text-centered">31.00</td>
<td class="has-text-centered">21.15</td>
<td class="has-text-centered">23.28</td>
<td class="has-text-centered">3.75</td>
<td class="has-text-centered">1.06</td>
<td class="has-text-centered">15.02</td>
<td class="has-text-centered">7.31</td>
<td class="has-text-centered">8.34</td>
<td class="has-text-centered">11.21</td>
<td class="has-text-centered">1.88</td>
<td class="has-text-centered">13.72</td>
<td class="has-text-centered">39.42</td>
<td class="has-text-centered">9.07</td>
<td class="column-border-left has-text-centered performance-strong">263.16</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="QwQ-32B-Preview" data-tooltip="Alibaba Cloud's 32 billion parameter QwenNext variant with optimizations for both general and specialized tasks. Features strong performance across various financial benchmarks with smaller parameter count.">QwQ-32B-Preview</td>
<td class="column-border-left has-text-centered">0.15</td>
<td class="has-text-centered">0.18</td>
<td class="has-text-centered">2.38</td>
<td class="has-text-centered">0.08</td>
<td class="has-text-centered">0.93</td>
<td class="has-text-centered">1.37</td>
<td class="has-text-centered">0.60</td>
<td class="has-text-centered">0.68</td>
<td class="has-text-centered">2.18</td>
<td class="has-text-centered">3.12</td>
<td class="has-text-centered">2.36</td>
<td class="has-text-centered">0.11</td>
<td class="has-text-centered">0.07</td>
<td class="has-text-centered">2.76</td>
<td class="has-text-centered">0.14</td>
<td class="has-text-centered">0.65</td>
<td class="has-text-centered">0.54</td>
<td class="has-text-centered">0.14</td>
<td class="has-text-centered">2.61</td>
<td class="has-text-centered">7.83</td>
<td class="has-text-centered">1.55</td>
<td class="column-border-left has-text-centered performance-medium">30.43</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="Jamba 1.5 Mini" data-tooltip="AI21 Labs' 52 billion parameter Mixture of Experts (MoE) model designed for efficient instruction following. Optimized for practical deployment with reduced computational requirements compared to larger variants.">Jamba 1.5 Mini</td>
<td class="column-border-left has-text-centered">0.02</td>
<td class="has-text-centered">0.03</td>
<td class="has-text-centered">0.30</td>
<td class="has-text-centered">0.02</td>
<td class="has-text-centered">0.23</td>
<td class="has-text-centered">0.22</td>
<td class="has-text-centered">0.10</td>
<td class="has-text-centered">0.08</td>
<td class="has-text-centered">0.44</td>
<td class="has-text-centered">0.55</td>
<td class="has-text-centered">0.51</td>
<td class="has-text-centered">0.02</td>
<td class="has-text-centered">0.01</td>
<td class="has-text-centered">0.49</td>
<td class="has-text-centered">0.05</td>
<td class="has-text-centered">0.10</td>
<td class="has-text-centered">0.07</td>
<td class="has-text-centered">0.02</td>
<td class="has-text-centered">0.25</td>
<td class="has-text-centered">0.72</td>
<td class="has-text-centered">0.26</td>
<td class="column-border-left has-text-centered performance-low">4.47</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="Jamba 1.5 Large" data-tooltip="AI21 Labs' premium 398 billion parameter Mixture of Experts (MoE) model designed for advanced reasoning. Features strong performance across complex tasks including financial domain problems.">Jamba 1.5 Large</td>
<td class="column-border-left has-text-centered">0.31</td>
<td class="has-text-centered">0.36</td>
<td class="has-text-centered">4.42</td>
<td class="has-text-centered">0.30</td>
<td class="has-text-centered">3.47</td>
<td class="has-text-centered">4.81</td>
<td class="has-text-centered">1.78</td>
<td class="has-text-centered">0.94</td>
<td class="has-text-centered">4.97</td>
<td class="has-text-centered">5.80</td>
<td class="has-text-centered">5.51</td>
<td class="has-text-centered">0.35</td>
<td class="has-text-centered">0.13</td>
<td class="has-text-centered">7.07</td>
<td class="has-text-centered">0.56</td>
<td class="has-text-centered">1.67</td>
<td class="has-text-centered">0.77</td>
<td class="has-text-centered">0.30</td>
<td class="has-text-centered">2.87</td>
<td class="has-text-centered">7.45</td>
<td class="has-text-centered">2.59</td>
<td class="column-border-left has-text-centered performance-strong">56.42</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="Claude 3.5 Sonnet" data-tooltip="Anthropic's advanced proprietary language model designed for complex reasoning and instruction-following tasks. Features strong performance across financial tasks with sophisticated reasoning capabilities.">Claude 3.5 Sonnet</td>
<td class="column-border-left has-text-centered">0.62</td>
<td class="has-text-centered">0.72</td>
<td class="has-text-centered">6.98</td>
<td class="has-text-centered">0.55</td>
<td class="has-text-centered">6.50</td>
<td class="has-text-centered">8.81</td>
<td class="has-text-centered">3.44</td>
<td class="has-text-centered">3.21</td>
<td class="has-text-centered">12.32</td>
<td class="has-text-centered">9.50</td>
<td class="has-text-centered">11.11</td>
<td class="has-text-centered">0.61</td>
<td class="has-text-centered">0.22</td>
<td class="has-text-centered">7.09</td>
<td class="has-text-centered">0.90</td>
<td class="has-text-centered">3.01</td>
<td class="has-text-centered">1.79</td>
<td class="has-text-centered">0.57</td>
<td class="has-text-centered">9.18</td>
<td class="has-text-centered">16.86</td>
<td class="has-text-centered">3.89</td>
<td class="column-border-left has-text-centered performance-strong">107.87</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="Claude 3 Haiku" data-tooltip="Anthropic's compact proprietary language model optimized for efficiency while maintaining strong capabilities. Designed for cost-effective deployment with reduced computational requirements compared to larger variants.">Claude 3 Haiku</td>
<td class="column-border-left has-text-centered">0.06</td>
<td class="has-text-centered">0.07</td>
<td class="has-text-centered">0.56</td>
<td class="has-text-centered">0.05</td>
<td class="has-text-centered">0.54</td>
<td class="has-text-centered">0.73</td>
<td class="has-text-centered">0.28</td>
<td class="has-text-centered">0.25</td>
<td class="has-text-centered">0.82</td>
<td class="has-text-centered">0.81</td>
<td class="has-text-centered">0.90</td>
<td class="has-text-centered">0.05</td>
<td class="has-text-centered">0.02</td>
<td class="has-text-centered">0.21</td>
<td class="has-text-centered">0.06</td>
<td class="has-text-centered">0.23</td>
<td class="has-text-centered">0.14</td>
<td class="has-text-centered">0.05</td>
<td class="has-text-centered">0.64</td>
<td class="has-text-centered">1.28</td>
<td class="has-text-centered">0.32</td>
<td class="column-border-left has-text-centered performance-low">8.07</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="Cohere Command R 7B" data-tooltip="Cohere's compact 7 billion parameter instruction-tuned model optimized for practical applications. Focused on efficiency with minimal computational requirements while maintaining reasonable capabilities.">Cohere Command R 7B</td>
<td class="column-border-left has-text-centered">0.01</td>
<td class="has-text-centered">0.01</td>
<td class="has-text-centered">0.08</td>
<td class="has-text-centered">0.00</td>
<td class="has-text-centered">0.07</td>
<td class="has-text-centered">0.09</td>
<td class="has-text-centered">0.04</td>
<td class="has-text-centered">0.03</td>
<td class="has-text-centered">0.11</td>
<td class="has-text-centered">0.11</td>
<td class="has-text-centered">0.10</td>
<td class="has-text-centered">0.01</td>
<td class="has-text-centered">0.00</td>
<td class="has-text-centered">0.08</td>
<td class="has-text-centered">0.01</td>
<td class="has-text-centered">0.03</td>
<td class="has-text-centered">0.01</td>
<td class="has-text-centered">0.01</td>
<td class="has-text-centered">0.08</td>
<td class="has-text-centered">0.19</td>
<td class="has-text-centered">0.05</td>
<td class="column-border-left has-text-centered performance-low">1.09</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="Cohere Command R +" data-tooltip="Cohere's premium 104 billion parameter dense language model optimized for advanced tasks. Features strong instruction-following capabilities with comprehensive knowledge across domains including finance.">Cohere Command R +</td>
<td class="column-border-left has-text-centered">0.41</td>
<td class="has-text-centered">0.45</td>
<td class="has-text-centered">5.40</td>
<td class="has-text-centered">0.35</td>
<td class="has-text-centered">4.41</td>
<td class="has-text-centered">4.00</td>
<td class="has-text-centered">2.30</td>
<td class="has-text-centered">0.93</td>
<td class="has-text-centered">3.87</td>
<td class="has-text-centered">7.03</td>
<td class="has-text-centered">7.21</td>
<td class="has-text-centered">0.43</td>
<td class="has-text-centered">0.12</td>
<td class="has-text-centered">5.55</td>
<td class="has-text-centered">0.48</td>
<td class="has-text-centered">1.69</td>
<td class="has-text-centered">0.97</td>
<td class="has-text-centered">0.42</td>
<td class="has-text-centered">4.59</td>
<td class="has-text-centered">10.09</td>
<td class="has-text-centered">3.24</td>
<td class="column-border-left has-text-centered performance-strong">63.95</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="Google Gemini 1.5 Pro" data-tooltip="Google's advanced proprietary multimodal model designed for complex reasoning and instruction-following tasks. Features strong performance across financial domains with advanced reasoning capabilities.">Google Gemini 1.5 Pro</td>
<td class="column-border-left has-text-centered">0.23</td>
<td class="has-text-centered">0.21</td>
<td class="has-text-centered">2.26</td>
<td class="has-text-centered">0.18</td>
<td class="has-text-centered">2.20</td>
<td class="has-text-centered">2.78</td>
<td class="has-text-centered">1.02</td>
<td class="has-text-centered">0.49</td>
<td class="has-text-centered">2.27</td>
<td class="has-text-centered">3.45</td>
<td class="has-text-centered">2.70</td>
<td class="has-text-centered">0.21</td>
<td class="has-text-centered">0.07</td>
<td class="has-text-centered">2.65</td>
<td class="has-text-centered">0.25</td>
<td class="has-text-centered">0.87</td>
<td class="has-text-centered">0.58</td>
<td class="has-text-centered">0.21</td>
<td class="has-text-centered">2.13</td>
<td class="has-text-centered">5.78</td>
<td class="has-text-centered">1.62</td>
<td class="column-border-left has-text-centered performance-medium">32.16</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="OpenAI GPT-4o" data-tooltip="OpenAI's flagship multimodal model optimized for a balance of quality and speed. Features strong performance across diverse tasks with capabilities for complex financial reasoning and instruction following.">OpenAI gpt-4o</td>
<td class="column-border-left has-text-centered">0.35</td>
<td class="has-text-centered">0.41</td>
<td class="has-text-centered">4.99</td>
<td class="has-text-centered">0.32</td>
<td class="has-text-centered">4.45</td>
<td class="has-text-centered">5.33</td>
<td class="has-text-centered">1.55</td>
<td class="has-text-centered">1.21</td>
<td class="has-text-centered">5.77</td>
<td class="has-text-centered">6.57</td>
<td class="has-text-centered">5.00</td>
<td class="has-text-centered">0.35</td>
<td class="has-text-centered">0.14</td>
<td class="has-text-centered">4.85</td>
<td class="has-text-centered">0.44</td>
<td class="has-text-centered">1.94</td>
<td class="has-text-centered">0.96</td>
<td class="has-text-centered">0.34</td>
<td class="has-text-centered">4.95</td>
<td class="has-text-centered">10.36</td>
<td class="has-text-centered">3.24</td>
<td class="column-border-left has-text-centered performance-strong">63.52</td>
</tr>
<tr>
<td class="tooltip-trigger tooltip-right" data-title="OpenAI o1-mini" data-tooltip="OpenAI's compact model from the o1 series, designed as a more efficient alternative to larger models. Features strong performance relative to its size with capabilities for complex financial tasks.">OpenAI o1-mini</td>
<td class="column-border-left has-text-centered">0.90</td>
<td class="has-text-centered">0.90</td>
<td class="has-text-centered">5.25</td>
<td class="has-text-centered">0.73</td>
<td class="has-text-centered">9.70</td>
<td class="has-text-centered">12.20</td>
<td class="has-text-centered">3.27</td>
<td class="has-text-centered">4.89</td>
<td class="has-text-centered">13.60</td>
<td class="has-text-centered">1.29</td>
<td class="has-text-centered">9.29</td>
<td class="has-text-centered">2.56</td>
<td class="has-text-centered">0.75</td>
<td class="has-text-centered">3.18</td>
<td class="has-text-centered">2.92</td>
<td class="has-text-centered">1.91</td>
<td class="has-text-centered">6.39</td>
<td class="has-text-centered">0.92</td>
<td class="has-text-centered">6.97</td>
<td class="has-text-centered">15.71</td>
<td class="has-text-centered">1.42</td>
<td class="column-border-left has-text-centered performance-strong">104.73</td>
</tr>
</tbody>
</table>
<style>
/* Ensure model tooltips appear on the right with arrow pointing left */
.tooltip-trigger.tooltip-right:hover::after {
left: 100% !important;
right: auto !important;
margin-left: 10px !important;
margin-right: 0 !important;
}
/* Fix for cost analysis tab model tooltips */
#cost-analysis td.tooltip-trigger:hover::after {
position: absolute !important;
left: 100% !important;
right: auto !important;
top: 50% !important;
transform: translateY(-50%) !important;
margin-left: 10px !important;
margin-right: 0 !important;
margin-top: 0 !important;
z-index: 99 !important;
}
#cost-analysis td.tooltip-trigger:hover::before {
position: absolute !important;
left: 100% !important;
right: auto !important;
top: 50% !important;
transform: translateY(-50%) !important;
border-left: 6px solid rgba(0, 0, 0, 0.8) !important;
border-right: none !important;
border-top: 6px solid transparent !important;
border-bottom: 6px solid transparent !important;
margin-top: 0 !important;
margin-left: 0 !important;
z-index: 100 !important;
}
/* Ensure the model tooltips are properly centered in the cost analysis tab */
.model-tooltip {
position: fixed !important;
transform: translateY(-50%) !important;
margin-top: 0 !important;
z-index: 1000 !important;
max-height: 400px !important;
overflow-y: auto !important;
}
/* Make sure arrows are properly centered */
.model-tooltip > div {
transform: translateY(-50%) !important;
z-index: 1001 !important;
}
/* Ensure arrows are visible and properly sized */
.model-tooltip > div,
#cost-analysis td.tooltip-trigger:first-child:hover::before {
border-width: 8px !important;
}
/* Force all model tooltips in cost analysis to have tooltip-right behavior */
#cost-analysis td:first-child {
position: relative;
}
/* More specific selectors to ensure model tooltips in cost analysis tab are positioned correctly */
#cost-analysis td.tooltip-trigger:first-child:hover::after {
left: 100% !important;
top: 50% !important;
bottom: auto !important;
transform: translateY(-50%) !important;
margin-top: 0 !important;
margin-left: 15px !important;
display: block !important;
}
#cost-analysis td.tooltip-trigger:first-child:hover::before {
content: '' !important;
display: block !important;
position: absolute !important;
top: 50% !important;
left: 100% !important;
right: auto !important;
bottom: auto !important;
transform: translateY(-50%) !important;
margin-left: 0 !important;
margin-top: 0 !important;
width: 0 !important;
height: 0 !important;
border-top: 10px solid transparent !important;
border-bottom: 10px solid transparent !important;
border-right: none !important;
border-left: 10px solid black !important;
z-index: 9999 !important;
pointer-events: none !important;
}
/* Add an extra arrow to each model cell specifically for visual clarity */
#cost-analysis td:first-child.tooltip-trigger:after {
content: "►" !important;
position: absolute !important;
right: 10px !important;
color: gray !important;
font-size: 12px !important;
}
</style>
<div class="performance-legend">
<div class="performance-legend-item performance-low">Low Cost ($0-$10)</div>
<div class="performance-legend-item performance-medium">Medium Cost ($10-$35)</div>
<div class="performance-legend-item performance-strong">High Cost ($35-$70)</div>
<div class="performance-legend-item" style="background-color: #1f93ff; color: white;">Very High Cost ($70+)</div>
</div>
<div class="content is-small mt-4">
<p><strong>Note:</strong> All costs are in USD and represent the expense to run the model on each specific dataset. Colors indicate cost tiers on the total cost, with darker blue representing higher costs. For cost-efficiency analysis, consider comparing these costs with the corresponding performance metrics in other tabs.</p>
</div>
</div>
</div>
</div>
</div>
</div>
</section>
<!-- </section>
</div> -->
<section class="section">
<div class="container">
<!-- Model Performance Highlights -->
<div class="card mb-5">
<div class="card-header">
<p class="card-header-title">
<span class="icon mr-2"><i class="fas fa-medal"></i></span>
Model Performance Highlights
</p>
</div>
<div class="card-content">
<div class="content">
<p class="has-text-weight-bold mb-4 has-text-centered">🏆 Top Performing Models</p>
<div class="columns is-centered is-multiline">
<!-- DeepSeek R1 -->
<div class="column is-4">
<div class="is-flex is-flex-direction-column is-align-items-center">
<figure class="image is-128x128 mb-3">
<img src="static/images/deepseek_logo.png" alt="DeepSeek R1 Logo">
</figure>
<p class="is-size-4 has-text-weight-semibold mb-2">DeepSeek R1</p>
<span class="icon is-large has-text-warning"><i class="fas fa-trophy fa-2x"></i></span>
</div>
</div>
<!-- OpenAI o1-mini -->
<div class="column is-4">
<div class="is-flex is-flex-direction-column is-align-items-center">
<figure class="image is-128x128 mb-3">
<img src="static/images/openai_logo.png" alt="OpenAI Logo">
</figure>
<p class="is-size-4 has-text-weight-semibold mb-2">OpenAI o1-mini</p>
<span class="icon is-large has-text-grey"><i class="fas fa-trophy fa-2x"></i></span>
</div>
</div>
<!-- Claude 3.5 Sonnet -->
<div class="column is-4">
<div class="is-flex is-flex-direction-column is-align-items-center">
<figure class="image is-128x128 mb-3">
<img src="static/images/claude_logo.png" alt="Claude 3.5 Sonnet Logo">
</figure>
<p class="is-size-4 has-text-weight-semibold mb-2">Claude 3.5 Sonnet</p>
<span class="icon is-large has-text-bronze"><i class="fas fa-trophy fa-2x"></i></span>
</div>
</div>
</div>
<hr>
<p class="has-text-weight-bold mb-3"><span class="icon has-text-primary"><i class="fa-solid fa-magnifying-glass"></i></span> Key Insights from Model Analysis</p>
<div class="notification is-info is-light py-3 px-4">
<p><strong>🏆 No single dominant model:</strong> DeepSeek R1 leads in complex multi-step QA, while Claude 3.5 excels in sentiment tasks. GPT-4o is strong in classification and summarization.</p>
<p><strong>⚖️ Inconsistent scaling:</strong> Larger models don’t always outperform smaller ones—DeepSeek R1 trails in summarization despite excelling in QA.</p>
<p><strong>🛠️ Open-weight models:</strong> Many open-weight models like DeepSeek-V3 and Llama 3.1 70B offer competitive performance while being cost-effective.</p>
<p><strong>💰 Cost-performance disparities:</strong> Running DeepSeek R1 can cost up to <strong>$260</strong> per million tokens, while Claude 3.5 Sonnet and o1-mini cost around <strong>$105</strong>, and Meta’s Llama 3.1 8B only <strong>$4</strong>.</p>
<p><strong>📈 Numeric reasoning challenges:</strong> Even the best models struggle with financial numeric reasoning tasks, achieving low F1 scores (<strong>≤ 0.06</strong>).</p>
<p><strong>🔢 Step-by-step deductions:</strong> Multi-turn financial QA (e.g., ConvFinQA) significantly reduces model accuracy due to complex dependencies.</p>
</div>
</div>
</div>
</div>
<!-- Error Analysis & Key Findings -->
<div class="card">
<div class="card-header">
<p class="card-header-title">
<span class="icon mr-2"><i class="fas fa-exclamation-triangle"></i></span>
Error Analysis & Key Findings
</p>
</div>
<div class="card-content">
<div class="content">
<p class="mb-4">Common challenges and limitations identified in our evaluations:</p>
<!-- Individual Error Categories -->
<div class="error-category mb-4">
<p class="has-text-weight-bold mb-1">Concerns regarding outdated models</p>
<div class="notification is-danger is-light py-2 px-3">
<p class="is-size-7 mb-0"><strong>LLama 2 13B Chat</strong> produces trivial or empty responses, possibly due to misalignment during fine-tuning.</p>
</div>
</div>
<div class="error-category mb-4">
<p class="has-text-weight-bold mb-1">Numeric Regression Issues</p>
<div class="notification is-danger is-light py-2 px-3">
<p class="is-size-7 mb-0">LMs struggle with precision and rounding in continuous-valued regressions (e.g., financial percentages). Post-hoc normalization is needed.</p>
</div>
</div>
<div class="error-category mb-4">
<p class="has-text-weight-bold mb-1">Data Contamination</p>
<div class="notification is-danger is-light py-2 px-3">
<p class="is-size-7 mb-0">Overlap between public financial datasets and pretraining corpora can inflate zero-shot performance, requiring time-split test sets.</p>
</div>
</div>
<div class="error-category mb-4">
<p class="has-text-weight-bold mb-1">Challenges in Causal Classification</p>
<div class="notification is-danger is-light py-2 px-3">
<p class="is-size-7 mb-0">Most models struggle with financial causal reasoning, requiring structured knowledge bases or explicit symbolic reasoning.</p>
</div>
</div>
<div class="error-category mb-4">
<p class="has-text-weight-bold mb-1">Language Drift</p>
<div class="notification is-warning is-light py-2 px-3">
<p class="is-size-7 mb-0"><strong>Qwen 2 72B</strong> exhibits unintended shifts to Chinese output in English summarization tasks, indicating strong pretraining priors.</p>
</div>
</div>
<div class="error-category mb-4">
<p class="has-text-weight-bold mb-1">Summarization Nuances</p>
<div class="notification is-warning is-light py-2 px-3">
<p class="is-size-7 mb-0">Models achieve high BERTScores (~80-82%) on extractive summarization but suffer on abstractive tasks, especially in finance-specific jargon.</p>
</div>
</div>
<div class="error-category mb-4">
<p class="has-text-weight-bold mb-1">Prompt Design Limitations</p>
<div class="notification is-warning is-light py-2 px-3">
<p class="is-size-7 mb-0">Prompts tuned on <strong>Llama 3 8B</strong> may not generalize across models, leading to inconsistencies in label generation (e.g., minor syntactic variations).</p>
</div>
</div>
<div class="error-category mb-4">
<p class="has-text-weight-bold mb-1">Differences in QA Datasets</p>
<div class="notification is-warning is-light py-2 px-3">
<p class="is-size-7 mb-0"><strong>ConvFinQA</strong> consistently underperforms compared to <strong>FinQA</strong> due to its multi-turn dialogue complexity.</p>
</div>
</div>
<div class="error-category mb-4">
<p class="has-text-weight-bold mb-1">Efficiency and Cost Considerations</p>
<div class="notification is-warning is-light py-2 px-3">
<p class="is-size-7 mb-0">Inference costs vary by up to <strong>2×</strong> among similarly sized models, requiring a balance between performance and resource usage.</p>
</div>
</div>
</div>
</div>
</div>
</div>
</section>
<section id="contributions" class="section content-section">
<div class="container is-max-desktop">
<!-- Contributions -->
<div class="columns is-centered">
<div class="column is-full">
<h2 class="title is-3 section-title has-text-centered">Contributions & Future Work</h2>
<div class="content">
<!-- Contributions Overview -->
<div class="notification is-info is-light has-text-centered mb-5">
<p class="is-size-5 has-text-weight-semibold">
Our work introduces a standardized, large-scale, and holistic evaluation framework for financial language models.
</p>
</div>
<!-- Contributions -->
<div class="box has-background-white-ter mb-5">
<h4 class="title is-4 has-text-centered mb-4">Key Contributions</h4>
<div class="columns is-multiline is-centered">
<!-- Contribution 1 -->
<div class="column is-6">
<div class="media">
<div class="media-left">
<span class="icon has-text-primary is-large">
<i class="fas fa-cogs fa-lg"></i>
</span>
</div>
<div class="media-content">
<p class="has-text-weight-semibold">Standardized Evaluation Framework</p>
<p class="is-size-7">We introduce an open-source, modular benchmarking suite for systematic LM evaluations on core financial NLP tasks.</p>
</div>
</div>
</div>
<!-- Contribution 2 -->
<div class="column is-6">
<div class="media">
<div class="media-left">
<span class="icon has-text-primary is-large">
<i class="fas fa-chart-line fa-lg"></i>
</span>
</div>
<div class="media-content">
<p class="has-text-weight-semibold">Large-Scale Model Assessment</p>
<p class="is-size-7">We benchmark 23 foundation LMs—open-weight and proprietary—across 20 financial tasks, revealing performance-cost trade-offs.</p>
</div>
</div>
</div>
<!-- Contribution 3 -->
<div class="column is-6">
<div class="media">
<div class="media-left">
<span class="icon has-text-primary is-large">
<i class="fas fa-database fa-lg"></i>
</span>
</div>
<div class="media-content">
<p class="has-text-weight-semibold">Holistic Dataset Taxonomy</p>
<p class="is-size-7">We establish a structured dataset taxonomy, categorizing financial NLP tasks based on domain, data format, and linguistic complexity.</p>
</div>
</div>
</div>
<!-- Contribution 4 -->
<div class="column is-6">
<div class="media">
<div class="media-left">
<span class="icon has-text-primary is-large">
<i class="fas fa-users fa-lg"></i>
</span>
</div>
<div class="media-content">
<p class="has-text-weight-semibold">Living Benchmark & Open Collaboration</p>
<p class="is-size-7">We introduce a continuously updated leaderboard, inviting researchers to contribute new datasets and evaluation results.</p>
</div>
</div>
</div>
<!-- Contribution 5 -->
<div class="column is-6">
<div class="media">
<div class="media-left">
<span class="icon has-text-primary is-large">
<i class="fas fa-balance-scale fa-lg"></i>
</span>
</div>
<div class="media-content">
<p class="has-text-weight-semibold">Error Analysis & Cost-Performance Insights</p>
<p class="is-size-7">We analyze systematic model errors and quantify cost-performance trade-offs for informed deployment in real-world applications.</p>
</div>
</div>
</div>
<!-- Contribution 6 -->
<div class="column is-6">
<div class="media">
<div class="media-left">
<span class="icon has-text-primary is-large">
<i class="fas fa-code-branch fa-lg"></i>
</span>
</div>
<div class="media-content">
<p class="has-text-weight-semibold">Open-Source Implementation</p>
<p class="is-size-7">We release a fully open-source framework, enabling the research community to extend and refine financial LM evaluation methodologies.</p>
</div>
</div>
</div>
</div>
</div>
<!-- Limitations & Future Work -->
<div class="columns is-multiline">
<!-- Limitations -->
<div class="column is-6">
<div class="card h-100">
<div class="card-header">
<p class="card-header-title">
<span class="icon mr-2"><i class="fas fa-exclamation-circle"></i></span>
Limitations
</p>
</div>
<div class="card-content">
<div class="content">
<p class="mb-3">
While our benchmark provides valuable insights, several limitations must be acknowledged:
</p>
<div class="notification is-danger is-light py-2 px-3 mb-3">
<p class="has-text-weight-bold mb-1">❌ Data Contamination Risks</p>
<p class="is-size-7 mb-0">Benchmark testing data may overlap with model pretraining corpora, leading to artificially inflated performance. We actively work on novel datasets to mitigate these risks.</p>
</div>
<div class="notification is-warning is-light py-2 px-3 mb-3">
<p class="has-text-weight-bold mb-1">⚠️ Dataset Size & Diversity</p>
<p class="is-size-7 mb-0">Our dataset scope is limited, affecting model generalization across diverse financial domains and languages.</p>
</div>
<div class="notification is-warning is-light py-2 px-3 mb-3">
<p class="has-text-weight-bold mb-1">⚠️ Zero-Shot Focus</p>
<p class="is-size-7 mb-0">Due to budget constraints, our evaluations rely on zero-shot learning only, without fine-tuning or few-shot prompting.</p>
</div>
<div class="notification is-warning is-light py-2 px-3 mb-3">
<p class="has-text-weight-bold mb-1">⚠️ Limited Adaptation Strategies</p>
<p class="is-size-7 mb-0">We do not explore chain-of-thought reasoning or advanced prompting, though these techniques are known to improve model performance.</p>
</div>
<div class="notification is-info is-light py-2 px-3 mb-3">
<p class="has-text-weight-bold mb-1">ℹ️ English Language Bias</p>
<p class="is-size-7 mb-0">The benchmark primarily focuses on English due to the availability of financial datasets, limiting insights into multilingual model performance.</p>
</div>
<div class="notification is-info is-light py-2 px-3 mb-3">
<p class="has-text-weight-bold mb-1">ℹ️ Real-World Complex Tasks</p>
<p class="is-size-7 mb-0">Existing tasks do not fully capture the dynamic and evolving nature of financial markets, requiring ongoing dataset expansion.</p>
</div>
<p class="is-italic is-size-7 mt-4">
Recognizing these limitations is essential for improving future financial NLP benchmarks. Our ongoing work aims to address these challenges through dataset refinement, broader task coverage, and multilingual support.
</p>
</div>
</div>
</div>
</div>
<!-- Future Work -->
<div class="column is-6">
<div class="card h-100">
<div class="card-header">
<p class="card-header-title">
<span class="icon mr-2"><i class="fas fa-lightbulb"></i></span>
Future Work
</p>
</div>
<div class="card-content">
<div class="content">
<p class="mb-3">
To strengthen the robustness and adaptability of our framework, we advocate for open collaboration within the research community
and propose the following future directions to expand its capabilities:
</p>
<div class="notification is-info is-light py-2 px-3 mb-3">
<p class="has-text-weight-bold mb-1">🌍 Multilingual Expansion</p>
<p class="is-size-7 mb-0">Extending benchmarks beyond English to include multilingual financial datasets and evaluations.</p>
</div>
<div class="notification is-info is-light py-2 px-3 mb-3">
<p class="has-text-weight-bold mb-1">🧠 Few-Shot & Chain-of-Thought</p>
<p class="is-size-7 mb-0">Investigating in-context learning techniques such as few-shot, chain-of-thought, and retrieval-augmented generation (RAG).</p>
</div>
<div class="notification is-info is-light py-2 px-3 mb-3">
<p class="has-text-weight-bold mb-1">⚙️ Domain-Adaptive Training</p>
<p class="is-size-7 mb-0">Evaluating fine-tuning strategies to enhance model understanding of financial-specific terminology and reasoning.</p>
</div>
<div class="notification is-info is-light py-2 px-3 mb-3">
<p class="has-text-weight-bold mb-1">📊 Expanded Dataset Coverage</p>
<p class="is-size-7 mb-0">Curating datasets from underrepresented financial sectors such as insurance, derivatives, and central banking.</p>
</div>
<div class="notification is-info is-light py-2 px-3 mb-3">
<p class="has-text-weight-bold mb-1">⚖️ Efficiency & Cost Benchmarking</p>
<p class="is-size-7 mb-0">Developing detailed trade-off analyses between accuracy, latency, and cost to optimize real-world usability.</p>
</div>
<div class="notification is-info is-light py-2 px-3 mb-3">
<p class="has-text-weight-bold mb-1">📈 Advanced Evaluation Metrics</p>
<p class="is-size-7 mb-0">Moving beyond traditional accuracy metrics by incorporating trustworthiness, robustness, and interpretability measures.</p>
</div>
<p class="is-italic is-size-7 mt-4">
These improvements will enable more accurate and fair comparisons of financial language models,
fostering greater transparency, reproducibility, and real-world applicability.
</p>
</div>
</div>
</div>
</div>
</div>
<!--/ Contributions -->
</div>
</section>
<section id="framework" class="section content-section">
<div class="container is-max-desktop">
<!-- Framework Overview -->
<div class="columns is-centered">
<div class="column is-full">
<h2 class="title is-3 section-title has-text-centered">Framework & Resources</h2>
<div class="box has-background-white-ter has-text-centered p-5 mb-5">
<span class="icon is-large mb-4">
<i class="fas fa-fire flame-icon fa-3x"></i>
</span>
<h3 class="title is-4"><span class="flame">FLaME</span> Framework</h3>
<p class="subtitle">An open-source, modular benchmarking suite for evaluating financial language models.</p>
<div class="buttons is-centered mt-4">
<a href="https://github.com/gtfintechlab/FLaME" target="_blank" class="button is-primary">
<span class="icon"><i class="fab fa-github"></i></span>
<span>GitHub Repository</span>
</a>
<a href="https://huggingface.co/gtfintechlab/" target="_blank" class="button is-info">
<span class="icon"><img src="./static/images/huggingface_logo.svg" alt="HuggingFace" width="20" height="20"></span>
<span>Hugging Face Space</span>
</a>
</div>
</div>
</div>
</div>
<!-- Features & Getting Started -->
<div class="columns is-multiline">
<!-- 🛠 Framework Features -->
<div class="column is-6">
<div class="card h-100">
<div class="card-header">
<p class="card-header-title">
<span class="icon mr-2"><i class="fas fa-cogs"></i></span>
Framework Features
</p>
</div>
<div class="card-content">
<div class="content">
<div class="feature-list">
<div class="feature-item mb-3">
<p class="has-text-weight-bold mb-1">
<span class="icon has-text-primary"><i class="fas fa-check"></i></span> 🔧 Standardized Pipelines
</p>
<p class="is-size-7 ml-4">Pre-built evaluation pipelines for key financial NLP tasks.</p>
</div>
<div class="feature-item mb-3">
<p class="has-text-weight-bold mb-1">
<span class="icon has-text-primary"><i class="fas fa-check"></i></span> ⚙️ Customizable Assessments
</p>
<p class="is-size-7 ml-4">Supports easy model-to-model and dataset-to-dataset comparisons.</p>
</div>
<div class="feature-item mb-3">
<p class="has-text-weight-bold mb-1">
<span class="icon has-text-primary"><i class="fas fa-check"></i></span> Reproducible Benchmarking
</p>
<p class="is-size-7 ml-4">Ensures consistent evaluation metrics and transparent methodology.</p>
</div>
<div class="feature-item">
<p class="has-text-weight-bold mb-1">
<span class="icon has-text-primary"><i class="fas fa-check"></i></span> 🔗 Extensible Architecture
</p>
<p class="is-size-7 ml-4">Easily integrates new tasks, datasets, and evaluation modules.</p>
</div>
</div>
</div>
</div>
</div>
</div>
<!-- 🚀 Getting Started -->
<div class="column is-6">
<div class="card h-100">
<div class="card-header">
<p class="card-header-title">
<span class="icon mr-2"><i class="fas fa-rocket"></i></span>
Getting Started
</p>
</div>
<div class="card-content">
<div class="content">
<div class="steps">
<div class="step-item">
<div class="step-marker">1</div>
<div class="step-details">
<p class="step-title">Clone Repository</p>
<p class="is-size-7 has-text-grey">
<code>git clone https://github.com/gtfintechlab/FLaME.git</code>
</p>
</div>
</div>
<div class="step-item">
<div class="step-marker">2</div>
<div class="step-details">
<p class="step-title">Install Dependencies</p>
<p class="is-size-7 has-text-grey">
<code>pip install -r requirements.txt</code>
</p>
</div>
</div>
<div class="step-item">
<div class="step-marker">3</div>
<div class="step-details">
<p class="step-title">Configure Models</p>
<p class="is-size-7 has-text-grey">Edit <code>config.yaml</code> with API keys and model settings.</p>
</div>
</div>
<div class="step-item">
<div class="step-marker">4</div>
<div class="step-details">
<p class="step-title">Run Evaluations</p>
<p class="is-size-7 has-text-grey">
<code>python -m flame.run --task all</code>
</p>
</div>
</div>
</div>
<div class="notification is-info is-light mt-4">
<p class="mb-0">For detailed documentation, visit our <a href="https://github.com/gtfintechlab/FLaME" target="_blank">GitHub repository</a>.</p>
</div>
</div>
</div>
</div>
</div>
</div> <!-- End Columns -->
</div>
</section>
<section id="datasets" class="section content-section">
<div class="container is-max-desktop">
<h2 class="title is-3 section-title has-text-centered">Datasets & Domains</h2>
<div class="columns is-multiline">
<!-- 📊 Numerical Reasoning & Question Answering -->
<div class="column is-6">
<div class="dataset-category box">
<p class="has-text-weight-bold">
📊 Numerical Reasoning & Question Answering
</p>
<ul>
<li><strong>FinQA</strong> – Multi-step financial numerical reasoning.</li>
<li><strong>ConvFinQA</strong> – Conversational numerical reasoning.</li>
<li><strong>TaT-QA</strong> – Hybrid table-text question answering.</li>
</ul>
</div>
</div>
<!-- 📝 Text Summarization -->
<div class="column is-6">
<div class="dataset-category box">
<p class="has-text-weight-bold">
📝 Text Summarization
</p>
<ul>
<li><strong>ECTSum</strong> – Earnings call transcript summarization.</li>
<li><strong>EDTSum</strong> – Financial news summarization.</li>
</ul>
</div>
</div>
<!-- 🔎 Information Retrieval -->
<div class="column is-6">
<div class="dataset-category box">
<p class="has-text-weight-bold">
🔎 Information Retrieval
</p>
<ul>
<li><strong>FiNER-ORD</strong> – Named entity recognition for financial documents.</li>
<li><strong>FinEntity</strong> – Entity-based sentiment classification.</li>
<li><strong>Financial Numeric Extreme Labeling (FNXL)</strong> – Automated numeral annotation in financial reports.</li>
<li><strong>FinRED</strong> – Relation extraction in finance.</li>
<li><strong>REFinD</strong> – Relation extraction from SEC filings.</li>
</ul>
</div>
</div>
<!-- 😐 Sentiment Analysis -->
<div class="column is-6">
<div class="dataset-category box">
<p class="has-text-weight-bold">
😐 Sentiment Analysis
</p>
<ul>
<li><strong>FiQA (Task 1)</strong> – Aspect-based sentiment analysis.</li>
<li><strong>FiQA (Task 2)</strong> – Opinion-based financial QA.</li>
<li><strong>Financial Phrase Bank (FPB)</strong> – Market sentiment classification.</li>
<li><strong>SubjECTive-QA</strong> – Subjectivity detection in earnings call Q&A.</li>
</ul>
</div>
</div>
<!-- 🏷️ Text Classification -->
<div class="column is-6">
<div class="dataset-category box">
<p class="has-text-weight-bold">
🏷️ Text Classification
</p>
<ul>
<li><strong>Numerical Claim Detection</strong> – Fine-grained investor claim detection.</li>
<li><strong>News Headline Classification</strong> – Market sentiment and trend detection.</li>
<li><strong>FOMC Dataset</strong> – Hawkish-Dovish stance classification.</li>
<li><strong>Banking77</strong> – Fine-grained intent detection.</li>
<li><strong>FinBench</strong> – Financial risk classification (default, fraud, churn).</li>
</ul>
</div>
</div>
<!-- 🧠 Causal Analysis -->
<div class="column is-6">
<div class="dataset-category box">
<p class="has-text-weight-bold">
🧠 Causal Analysis
</p>
<ul>
<li><strong>FinCausal</strong> – Causal reasoning in financial news.</li>
<li><strong>FinCausal-SC</strong> – Cause-effect span extraction.</li>
</ul>
</div>
</div>
</div> <!-- End Columns -->
</div>
</section>
<section id="BibTeX" class="section content-section">
<div class="container is-max-desktop">
<h2 class="title is-3 section-title has-text-centered">Citation</h2>
<div class="bibtex-container">
<div class="notification is-primary is-light mb-4">
<p class="has-text-centered mb-0">
If you use <span class="flame">FLaME</span> in your research, please cite our paper:
</p>
</div>
<div class="box">
<pre><code>@article{flame2025,
author = {Glenn Matlin, Mika Okamoto, Huzaifa Pardawala, Yang Yang, Sudheer Chava},
title = {FLaME: Holistic Financial Language Model Evaluation},
year = {2025},
month = {February},
}</code></pre>
</div>
</div>
</div>
</section>
<footer class="footer">
<div class="container">
<div class="columns is-centered">
<div class="column is-8">
<div class="content has-text-centered">
<h4 class="has-text-white mb-4"><span class="flame">FLaME</span>: Financial Language Model Evaluation</h4>
<div class="footer-links mb-5">
<a class="icon-link mr-3" target="_blank" href="FLaME/FLaME.pdf" title="Download PDF">
<i class="fas fa-file-pdf fa-lg"></i>
</a>
<!-- <a class="icon-link mr-3" href="https://arxiv.org/abs/" target="_blank" title="View on arXiv">
<i class="ai ai-arxiv fa-lg"></i>
</a> -->
<a class="icon-link mr-3" href="https://github.com/gtfintechlab/FLaME" target="_blank" title="GitHub Repository">
<i class="fab fa-github fa-lg"></i>
</a>
<a class="icon-link" href="https://huggingface.co/gtfintechlab/" target="_blank" title="HuggingFace Space">
<img src="./static/images/huggingface_logo.svg" alt="HuggingFace" width="24" height="24">
</a>
</div>
<div class="institution-info mb-4">
<p class="has-text-white-ter">Georgia Institute of Technology</p>
</div>
<p class="has-text-white-ter is-size-7">
This website is licensed under a <a rel="license" target="_blank" class="has-text-weight-bold" href="http://creativecommons.org/licenses/by-sa/4.0/">Creative Commons Attribution-ShareAlike 4.0 International License</a>.
</p>
</div>
</div>
</div>
</div>
</footer>
<!-- JavaScript for mobile menu toggle -->
<script>
document.addEventListener('DOMContentLoaded', () => {
// Get all "navbar-burger" elements
const $navbarBurgers = Array.prototype.slice.call(document.querySelectorAll('.navbar-burger'), 0);
// Add a click event on each of them
$navbarBurgers.forEach( el => {
el.addEventListener('click', () => {
// Get the target from the "data-target" attribute
const target = el.dataset.target;
const $target = document.getElementById(target);
// Toggle the "is-active" class on both the "navbar-burger" and the "navbar-menu"
el.classList.toggle('is-active');
$target.classList.toggle('is-active');
});
});
});
// Add padding to top of page for fixed navbar
document.addEventListener('DOMContentLoaded', () => {
const navbar = document.querySelector('.navbar');
const navbarHeight = navbar.offsetHeight;
document.body.style.paddingTop = navbarHeight + 'px';
});
</script>
</body>
</html> |