Skip to content

B
book

PaddlePaddle / book

通知 16
Star 4
Fork 0
B
book
切换分支/标签
查找文件 普通视图历史永久链接Permalink
README.en.html 51.3 KB
Newer Older
M
Sample chapter roundup for now  
Mimee 已提交
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 
<!DOCTYPE html><html><head><meta charset="utf-8"><style>body {
  width: 45em;
  border: 1px solid #ddd;
  outline: 1300px solid #fff;
  margin: 16px auto;
}

body .markdown-body
{
  padding: 30px;
}

@font-face {
  font-family: fontawesome-mini;
  src: url(data:font/woff;charset=utf-8;base64,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) format('woff');
}

@font-face {
  font-family: octicons-anchor;
  src: url(data:font/woff;charset=utf-8;base64,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) format('woff');
}

.markdown-body {
  font-family: sans-serif;
  -ms-text-size-adjust: 100%;
  -webkit-text-size-adjust: 100%;
  color: #333333;
  overflow: hidden;
  font-family: "Helvetica Neue", Helvetica, "Segoe UI", Arial, freesans, sans-serif;
  font-size: 16px;
  line-height: 1.6;
  word-wrap: break-word;
}

.markdown-body a {
  background: transparent;
}

.markdown-body a:active,
.markdown-body a:hover {
  outline: 0;
}

.markdown-body b,
.markdown-body strong {
  font-weight: bold;
}

.markdown-body mark {
  background: #ff0;
  color: #000;
  font-style: italic;
  font-weight: bold;
}

.markdown-body sub,
.markdown-body sup {
  font-size: 75%;
  line-height: 0;
  position: relative;
  vertical-align: baseline;
}
.markdown-body sup {
  top: -0.5em;
}
.markdown-body sub {
  bottom: -0.25em;
}

.markdown-body h1 {
  font-size: 2em;
  margin: 0.67em 0;
}

.markdown-body img {
  border: 0;
}

.markdown-body hr {
  -moz-box-sizing: content-box;
  box-sizing: content-box;
  height: 0;
}

.markdown-body pre {
  overflow: auto;
}

.markdown-body code,
.markdown-body kbd,
.markdown-body pre,
.markdown-body samp {
  font-family: monospace, monospace;
  font-size: 1em;
}

.markdown-body input {
  color: inherit;
  font: inherit;
  margin: 0;
}

.markdown-body html input[disabled] {
  cursor: default;
}

.markdown-body input {
  line-height: normal;
}

.markdown-body input[type="checkbox"] {
  box-sizing: border-box;
  padding: 0;
}

.markdown-body table {
  border-collapse: collapse;
  border-spacing: 0;
}

.markdown-body td,
.markdown-body th {
  padding: 0;
}

.markdown-body .codehilitetable {
  border: 0;
  border-spacing: 0;
}

.markdown-body .codehilitetable tr {
  border: 0;
}

.markdown-body .codehilitetable pre,
.markdown-body .codehilitetable div.codehilite {
  margin: 0;
}

.markdown-body .linenos,
.markdown-body .code,
.markdown-body .codehilitetable td {
  border: 0;
  padding: 0;
}

.markdown-body td:not(.linenos) .linenodiv {
  padding: 0 !important;
}

.markdown-body .code {
  width: 100%;
}

.markdown-body .linenos div pre,
.markdown-body .linenodiv pre,
.markdown-body .linenodiv {
  border: 0;
  -webkit-border-radius: 0;
  -moz-border-radius: 0;
  border-radius: 0;
  -webkit-border-top-left-radius: 3px;
  -webkit-border-bottom-left-radius: 3px;
  -moz-border-radius-topleft: 3px;
  -moz-border-radius-bottomleft: 3px;
  border-top-left-radius: 3px;
  border-bottom-left-radius: 3px;
}

.markdown-body .code div pre,
.markdown-body .code div {
  border: 0;
  -webkit-border-radius: 0;
  -moz-border-radius: 0;
  border-radius: 0;
  -webkit-border-top-right-radius: 3px;
  -webkit-border-bottom-right-radius: 3px;
  -moz-border-radius-topright: 3px;
  -moz-border-radius-bottomright: 3px;
  border-top-right-radius: 3px;
  border-bottom-right-radius: 3px;
}

.markdown-body * {
  -moz-box-sizing: border-box;
  box-sizing: border-box;
}

.markdown-body input {
  font: 13px Helvetica, arial, freesans, clean, sans-serif, "Segoe UI Emoji", "Segoe UI Symbol";
  line-height: 1.4;
}

.markdown-body a {
  color: #4183c4;
  text-decoration: none;
}

.markdown-body a:hover,
.markdown-body a:focus,
.markdown-body a:active {
  text-decoration: underline;
}

.markdown-body hr {
  height: 0;
  margin: 15px 0;
  overflow: hidden;
  background: transparent;
  border: 0;
  border-bottom: 1px solid #ddd;
}

.markdown-body hr:before,
.markdown-body hr:after {
  display: table;
  content: " ";
}

.markdown-body hr:after {
  clear: both;
}

.markdown-body h1,
.markdown-body h2,
.markdown-body h3,
.markdown-body h4,
.markdown-body h5,
.markdown-body h6 {
  margin-top: 15px;
  margin-bottom: 15px;
  line-height: 1.1;
}

.markdown-body h1 {
  font-size: 30px;
}

.markdown-body h2 {
  font-size: 21px;
}

.markdown-body h3 {
  font-size: 16px;
}

.markdown-body h4 {
  font-size: 14px;
}

.markdown-body h5 {
  font-size: 12px;
}

.markdown-body h6 {
  font-size: 11px;
}

.markdown-body blockquote {
  margin: 0;
}

.markdown-body ul,
.markdown-body ol {
  padding: 0;
  margin-top: 0;
  margin-bottom: 0;
}

.markdown-body ol ol,
.markdown-body ul ol {
  list-style-type: lower-roman;
}

.markdown-body ul ul ol,
.markdown-body ul ol ol,
.markdown-body ol ul ol,
.markdown-body ol ol ol {
  list-style-type: lower-alpha;
}

.markdown-body dd {
  margin-left: 0;
}

.markdown-body code,
.markdown-body pre,
.markdown-body samp {
  font-family: Consolas, "Liberation Mono", Menlo, Courier, monospace;
  font-size: 12px;
}

.markdown-body pre {
  margin-top: 0;
  margin-bottom: 0;
}

.markdown-body kbd {
  background-color: #e7e7e7;
  background-image: -moz-linear-gradient(#fefefe, #e7e7e7);
  background-image: -webkit-linear-gradient(#fefefe, #e7e7e7);
  background-image: linear-gradient(#fefefe, #e7e7e7);
  background-repeat: repeat-x;
  border-radius: 2px;
  border: 1px solid #cfcfcf;
  color: #000;
  padding: 3px 5px;
  line-height: 10px;
  font: 11px Consolas, "Liberation Mono", Menlo, Courier, monospace;
  display: inline-block;
}

.markdown-body>*:first-child {
  margin-top: 0 !important;
}

.markdown-body>*:last-child {
  margin-bottom: 0 !important;
}

.markdown-body .headeranchor-link {
  position: absolute;
  top: 0;
  bottom: 0;
  left: 0;
  display: block;
  padding-right: 6px;
  padding-left: 30px;
  margin-left: -30px;
}

.markdown-body .headeranchor-link:focus {
  outline: none;
}

.markdown-body h1,
.markdown-body h2,
.markdown-body h3,
.markdown-body h4,
.markdown-body h5,
.markdown-body h6 {
  position: relative;
  margin-top: 1em;
  margin-bottom: 16px;
  font-weight: bold;
  line-height: 1.4;
}

.markdown-body h1 .headeranchor,
.markdown-body h2 .headeranchor,
.markdown-body h3 .headeranchor,
.markdown-body h4 .headeranchor,
.markdown-body h5 .headeranchor,
.markdown-body h6 .headeranchor {
  display: none;
  color: #000;
  vertical-align: middle;
}

.markdown-body h1:hover .headeranchor-link,
.markdown-body h2:hover .headeranchor-link,
.markdown-body h3:hover .headeranchor-link,
.markdown-body h4:hover .headeranchor-link,
.markdown-body h5:hover .headeranchor-link,
.markdown-body h6:hover .headeranchor-link {
  height: 1em;
  padding-left: 8px;
  margin-left: -30px;
  line-height: 1;
  text-decoration: none;
}

.markdown-body h1:hover .headeranchor-link .headeranchor,
.markdown-body h2:hover .headeranchor-link .headeranchor,
.markdown-body h3:hover .headeranchor-link .headeranchor,
.markdown-body h4:hover .headeranchor-link .headeranchor,
.markdown-body h5:hover .headeranchor-link .headeranchor,
.markdown-body h6:hover .headeranchor-link .headeranchor {
  display: inline-block;
}

.markdown-body h1 {
  padding-bottom: 0.3em;
  font-size: 2.25em;
  line-height: 1.2;
  border-bottom: 1px solid #eee;
}

.markdown-body h2 {
  padding-bottom: 0.3em;
  font-size: 1.75em;
  line-height: 1.225;
  border-bottom: 1px solid #eee;
}

.markdown-body h3 {
  font-size: 1.5em;
  line-height: 1.43;
}

.markdown-body h4 {
  font-size: 1.25em;
}

.markdown-body h5 {
  font-size: 1em;
}

.markdown-body h6 {
  font-size: 1em;
  color: #777;
}

.markdown-body p,
.markdown-body blockquote,
.markdown-body ul,
.markdown-body ol,
.markdown-body dl,
.markdown-body table,
.markdown-body pre,
.markdown-body .admonition {
  margin-top: 0;
  margin-bottom: 16px;
}

.markdown-body hr {
  height: 4px;
  padding: 0;
  margin: 16px 0;
  background-color: #e7e7e7;
  border: 0 none;
}

.markdown-body ul,
.markdown-body ol {
  padding-left: 2em;
}

.markdown-body ul ul,
.markdown-body ul ol,
.markdown-body ol ol,
.markdown-body ol ul {
  margin-top: 0;
  margin-bottom: 0;
}

.markdown-body li>p {
  margin-top: 16px;
}

.markdown-body dl {
  padding: 0;
}

.markdown-body dl dt {
  padding: 0;
  margin-top: 16px;
  font-size: 1em;
  font-style: italic;
  font-weight: bold;
}

.markdown-body dl dd {
  padding: 0 16px;
  margin-bottom: 16px;
}

.markdown-body blockquote {
  padding: 0 15px;
  color: #777;
  border-left: 4px solid #ddd;
}

.markdown-body blockquote>:first-child {
  margin-top: 0;
}

.markdown-body blockquote>:last-child {
  margin-bottom: 0;
}

.markdown-body table {
  display: block;
  width: 100%;
  overflow: auto;
  word-break: normal;
  word-break: keep-all;
}

.markdown-body table th {
  font-weight: bold;
}

.markdown-body table th,
.markdown-body table td {
  padding: 6px 13px;
  border: 1px solid #ddd;
}

.markdown-body table tr {
  background-color: #fff;
  border-top: 1px solid #ccc;
}

.markdown-body table tr:nth-child(2n) {
  background-color: #f8f8f8;
}

.markdown-body img {
  max-width: 100%;
  -moz-box-sizing: border-box;
  box-sizing: border-box;
}

.markdown-body code,
.markdown-body samp {
  padding: 0;
  padding-top: 0.2em;
  padding-bottom: 0.2em;
  margin: 0;
  font-size: 85%;
  background-color: rgba(0,0,0,0.04);
  border-radius: 3px;
}

.markdown-body code:before,
.markdown-body code:after {
  letter-spacing: -0.2em;
  content: "\00a0";
}

.markdown-body pre>code {
  padding: 0;
  margin: 0;
  font-size: 100%;
  word-break: normal;
  white-space: pre;
  background: transparent;
  border: 0;
}

.markdown-body .codehilite {
  margin-bottom: 16px;
}

.markdown-body .codehilite pre,
.markdown-body pre {
  padding: 16px;
  overflow: auto;
  font-size: 85%;
  line-height: 1.45;
  background-color: #f7f7f7;
  border-radius: 3px;
}

.markdown-body .codehilite pre {
  margin-bottom: 0;
  word-break: normal;
}

.markdown-body pre {
  word-wrap: normal;
}

.markdown-body pre code {
  display: inline;
  max-width: initial;
  padding: 0;
  margin: 0;
  overflow: initial;
  line-height: inherit;
  word-wrap: normal;
  background-color: transparent;
  border: 0;
}

.markdown-body pre code:before,
.markdown-body pre code:after {
  content: normal;
}

/* Admonition */
.markdown-body .admonition {
  -webkit-border-radius: 3px;
  -moz-border-radius: 3px;
  position: relative;
  border-radius: 3px;
  border: 1px solid #e0e0e0;
  border-left: 6px solid #333;
  padding: 10px 10px 10px 30px;
}

.markdown-body .admonition table {
  color: #333;
}

.markdown-body .admonition p {
  padding: 0;
}

.markdown-body .admonition-title {
  font-weight: bold;
  margin: 0;
}

.markdown-body .admonition>.admonition-title {
  color: #333;
}

.markdown-body .attention>.admonition-title {
  color: #a6d796;
}

.markdown-body .caution>.admonition-title {
  color: #d7a796;
}

.markdown-body .hint>.admonition-title {
  color: #96c6d7;
}

.markdown-body .danger>.admonition-title {
  color: #c25f77;
}

.markdown-body .question>.admonition-title {
  color: #96a6d7;
}

.markdown-body .note>.admonition-title {
  color: #d7c896;
}

.markdown-body .admonition:before,
.markdown-body .attention:before,
.markdown-body .caution:before,
.markdown-body .hint:before,
.markdown-body .danger:before,
.markdown-body .question:before,
.markdown-body .note:before {
  font: normal normal 16px fontawesome-mini;
  -moz-osx-font-smoothing: grayscale;
  -webkit-user-select: none;
  -moz-user-select: none;
  -ms-user-select: none;
  user-select: none;
  line-height: 1.5;
  color: #333;
  position: absolute;
  left: 0;
  top: 0;
  padding-top: 10px;
  padding-left: 10px;
}

.markdown-body .admonition:before {
  content: "\f056\00a0";
  color: 333;
}

.markdown-body .attention:before {
  content: "\f058\00a0";
  color: #a6d796;
}

.markdown-body .caution:before {
  content: "\f06a\00a0";
  color: #d7a796;
}

.markdown-body .hint:before {
  content: "\f05a\00a0";
  color: #96c6d7;
}

.markdown-body .danger:before {
  content: "\f057\00a0";
  color: #c25f77;
}

.markdown-body .question:before {
  content: "\f059\00a0";
  color: #96a6d7;
}

.markdown-body .note:before {
  content: "\f040\00a0";
  color: #d7c896;
}

.markdown-body .admonition::after {
  content: normal;
}

.markdown-body .attention {
  border-left: 6px solid #a6d796;
}

.markdown-body .caution {
  border-left: 6px solid #d7a796;
}

.markdown-body .hint {
  border-left: 6px solid #96c6d7;
}

.markdown-body .danger {
  border-left: 6px solid #c25f77;
}

.markdown-body .question {
  border-left: 6px solid #96a6d7;
}

.markdown-body .note {
  border-left: 6px solid #d7c896;
}

.markdown-body .admonition>*:first-child {
  margin-top: 0 !important;
}

.markdown-body .admonition>*:last-child {
  margin-bottom: 0 !important;
}

/* progress bar*/
.markdown-body .progress {
  display: block;
  width: 300px;
  margin: 10px 0;
  height: 24px;
  -webkit-border-radius: 3px;
  -moz-border-radius: 3px;
  border-radius: 3px;
  background-color: #ededed;
  position: relative;
  box-shadow: inset -1px 1px 3px rgba(0, 0, 0, .1);
}

.markdown-body .progress-label {
  position: absolute;
  text-align: center;
  font-weight: bold;
  width: 100%; margin: 0;
  line-height: 24px;
  color: #333;
  text-shadow: 1px 1px 0 #fefefe, -1px -1px 0 #fefefe, -1px 1px 0 #fefefe, 1px -1px 0 #fefefe, 0 1px 0 #fefefe, 0 -1px 0 #fefefe, 1px 0 0 #fefefe, -1px 0 0 #fefefe, 1px 1px 2px #000;
  -webkit-font-smoothing: antialiased !important;
  white-space: nowrap;
  overflow: hidden;
}

.markdown-body .progress-bar {
  height: 24px;
  float: left;
  -webkit-border-radius: 3px;
  -moz-border-radius: 3px;
  border-radius: 3px;
  background-color: #96c6d7;
  box-shadow: inset 0 1px 0 rgba(255, 255, 255, .5), inset 0 -1px 0 rgba(0, 0, 0, .1);
  background-size: 30px 30px;
  background-image: -webkit-linear-gradient(
    135deg, rgba(255, 255, 255, .4) 27%,
    transparent 27%,
    transparent 52%, rgba(255, 255, 255, .4) 52%,
    rgba(255, 255, 255, .4) 77%,
    transparent 77%, transparent
  );
  background-image: -moz-linear-gradient(
    135deg,
    rgba(255, 255, 255, .4) 27%, transparent 27%,
    transparent 52%, rgba(255, 255, 255, .4) 52%,
    rgba(255, 255, 255, .4) 77%, transparent 77%,
    transparent
  );
  background-image: -ms-linear-gradient(
    135deg,
    rgba(255, 255, 255, .4) 27%, transparent 27%,
    transparent 52%, rgba(255, 255, 255, .4) 52%,
    rgba(255, 255, 255, .4) 77%, transparent 77%,
    transparent
  );
  background-image: -o-linear-gradient(
    135deg,
    rgba(255, 255, 255, .4) 27%, transparent 27%,
    transparent 52%, rgba(255, 255, 255, .4) 52%,
    rgba(255, 255, 255, .4) 77%, transparent 77%,
    transparent
  );
  background-image: linear-gradient(
    135deg,
    rgba(255, 255, 255, .4) 27%, transparent 27%,
    transparent 52%, rgba(255, 255, 255, .4) 52%,
    rgba(255, 255, 255, .4) 77%, transparent 77%,
    transparent
  );
}

.markdown-body .progress-100plus .progress-bar {
  background-color: #a6d796;
}

.markdown-body .progress-80plus .progress-bar {
  background-color: #c6d796;
}

.markdown-body .progress-60plus .progress-bar {
  background-color: #d7c896;
}

.markdown-body .progress-40plus .progress-bar {
  background-color: #d7a796;
}

.markdown-body .progress-20plus .progress-bar {
  background-color: #d796a6;
}

.markdown-body .progress-0plus .progress-bar {
  background-color: #c25f77;
}

.markdown-body .candystripe-animate .progress-bar{
  -webkit-animation: animate-stripes 3s linear infinite;
  -moz-animation: animate-stripes 3s linear infinite;
  animation: animate-stripes 3s linear infinite;
}

@-webkit-keyframes animate-stripes {
  0% {
    background-position: 0 0;
  }

  100% {
    background-position: 60px 0;
  }
}

@-moz-keyframes animate-stripes {
  0% {
    background-position: 0 0;
  }

  100% {
    background-position: 60px 0;
  }
}

@keyframes animate-stripes {
  0% {
    background-position: 0 0;
  }

  100% {
    background-position: 60px 0;
  }
}

.markdown-body .gloss .progress-bar {
  box-shadow:
    inset 0 4px 12px rgba(255, 255, 255, .7),
    inset 0 -12px 0 rgba(0, 0, 0, .05);
}

/* Multimarkdown Critic Blocks */
.markdown-body .critic_mark {
  background: #ff0;
}

.markdown-body .critic_delete {
  color: #c82829;
  text-decoration: line-through;
}

.markdown-body .critic_insert {
  color: #718c00 ;
  text-decoration: underline;
}

.markdown-body .critic_comment {
  color: #8e908c;
  font-style: italic;
}

.markdown-body .headeranchor {
  font: normal normal 16px octicons-anchor;
  line-height: 1;
  display: inline-block;
  text-decoration: none;
  -webkit-font-smoothing: antialiased;
  -moz-osx-font-smoothing: grayscale;
  -webkit-user-select: none;
  -moz-user-select: none;
  -ms-user-select: none;
  user-select: none;
}

.headeranchor:before {
  content: '\f05c';
}

.markdown-body .task-list-item {
  list-style-type: none;
}

.markdown-body .task-list-item+.task-list-item {
  margin-top: 3px;
}

.markdown-body .task-list-item input {
  margin: 0 4px 0.25em -20px;
  vertical-align: middle;
}

/* Media */
@media only screen and (min-width: 480px) {
  .markdown-body {
    font-size:14px;
  }
}

@media only screen and (min-width: 768px) {
  .markdown-body {
    font-size:16px;
  }
}

@media print {
  .markdown-body * {
    background: transparent !important;
    color: black !important;
    filter:none !important;
    -ms-filter: none !important;
  }

  .markdown-body {
    font-size:12pt;
    max-width:100%;
    outline:none;
    border: 0;
  }

  .markdown-body a,
  .markdown-body a:visited {
    text-decoration: underline;
  }

  .markdown-body .headeranchor-link {
    display: none;
  }

  .markdown-body a[href]:after {
    content: " (" attr(href) ")";
  }

  .markdown-body abbr[title]:after {
    content: " (" attr(title) ")";
  }

  .markdown-body .ir a:after,
  .markdown-body a[href^="javascript:"]:after,
  .markdown-body a[href^="#"]:after {
    content: "";
  }

  .markdown-body pre {
    white-space: pre;
    white-space: pre-wrap;
    word-wrap: break-word;
  }

  .markdown-body pre,
  .markdown-body blockquote {
    border: 1px solid #999;
    padding-right: 1em;
    page-break-inside: avoid;
  }

  .markdown-body .progress,
  .markdown-body .progress-bar {
    -moz-box-shadow: none;
    -webkit-box-shadow: none;
    box-shadow: none;
  }

  .markdown-body .progress {
    border: 1px solid #ddd;
  }

  .markdown-body .progress-bar {
    height: 22px;
    border-right: 1px solid #ddd;
  }

  .markdown-body tr,
  .markdown-body img {
    page-break-inside: avoid;
  }

  .markdown-body img {
    max-width: 100% !important;
  }

  .markdown-body p,
  .markdown-body h2,
  .markdown-body h3 {
    orphans: 3;
    widows: 3;
  }

  .markdown-body h2,
  .markdown-body h3 {
    page-break-after: avoid;
  }
}
</style><title>README.en</title></head><body><article class="markdown-body"><h1 id="sentiment-analysis"><a name="user-content-sentiment-analysis" href="#sentiment-analysis" class="headeranchor-link" aria-hidden="true"><span class="headeranchor"></span></a>Sentiment Analysis</h1>
<p>The source codes of this section can be located at <a href="https://github.com/PaddlePaddle/book/tree/develop/understand_sentiment">book/understand_sentiment</a>. First-time users may refer to PaddlePaddle for <a href="https://github.com/PaddlePaddle/Paddle/blob/develop/doc/getstarted/build_and_install/docker_install_en.rst">Installation guide</a>.</p>
<h2 id="background"><a name="user-content-background" href="#background" class="headeranchor-link" aria-hidden="true"><span class="headeranchor"></span></a>Background</h2>
<p>In natural language processing, sentiment analysis refers to determining the emotion expressed in a piece of text. The text can be a sentence, a paragraph, or a document. Emotion categorization can be binary &ndash; positive/negative or happy/sad &ndash; or in three classes &ndash; positive/neutral/negative. Sentiment analysis is applicable in a wide range of services, such as e-commerce sites like Amazon and Taobao, hospitality services like Airbnb and hotels.com, and movie rating sites like Rotten Tomatoes and IMDB. It can be used to gauge from the reviews how the customers feel about the product. Table 1 illustrates an example of sentiment analysis in movie reviews:</p>
<table>
<thead>
<tr>
<th>Movie Review</th>
<th>Category</th>
</tr>
</thead>
<tbody>
<tr>
<td>Best movie of Xiaogang Feng in recent years!</td>
<td>Positive</td>
</tr>
<tr>
<td>Pretty bad. Feels like a tv-series from a local TV-channel</td>
<td>Negative</td>
</tr>
<tr>
<td>Politically correct version of Taken &hellip; and boring as Heck</td>
<td>Negative</td>
</tr>
<tr>
<td>delightful, mesmerizing, and completely unexpected. The plot is nicely designed.</td>
<td>Positive</td>
</tr>
</tbody>
</table>
<p align="center">Table 1 Sentiment Analysis in Movie Reviews</p>

<p>In natural language processing, sentiment analysis can be categorized as a <strong>Text Classification problem</strong>, i.e., to categorize a piece of text to a specific class. It involves two related tasks: text representation and classification. Before the emergence of deep learning techniques, the mainstream methods for text representation include BOW (<em>bag of words</em>) and topic modeling, while the latter contain SVM (<em>support vector machine</em>) and LR (<em>logistic regression</em>).</p>
<p>The BOW model does not capture all the information in a piece of text, as it ignores syntax and grammar and just treats the text as a set of words. For example, “this movie is extremely bad“ and “boring, dull, and empty work” describe very similar semantic meaning, yet their BOW representations have with little similarity. Furthermore, “the movie is bad“ and “the movie is not bad“ have high similarity with BOW features, but they express completely opposite semantics.</p>
<p>This chapter introduces a deep learning model that handles these issues in BOW. Our model embeds texts into a low-dimensional space and takes word order into consideration. It is an end-to-end framework and it has large performance improvement over traditional methods [<a href="#Reference">1</a>].</p>
<h2 id="model-overview"><a name="user-content-model-overview" href="#model-overview" class="headeranchor-link" aria-hidden="true"><span class="headeranchor"></span></a>Model Overview</h2>
<p>The model we used in this chapter uses <strong>Convolutional Neural Networks</strong> (<strong>CNNs</strong>) and <strong>Recurrent Neural Networks</strong> (<strong>RNNs</strong>) with some specific extensions.</p>
<h3 id="convolutional-neural-networks-for-texts-cnn"><a name="user-content-convolutional-neural-networks-for-texts-cnn" href="#convolutional-neural-networks-for-texts-cnn" class="headeranchor-link" aria-hidden="true"><span class="headeranchor"></span></a>Convolutional Neural Networks for Texts (CNN)</h3>
<p><strong>Convolutional Neural Networks</strong> are frequently applied to data with grid-like topology such as two-dimensional images and one-dimensional texts. A CNN can extract multiple local features, combine them, and produce high-level abstractions, which correspond to semantic understanding. Empirically, CNN is shown to be efficient for image and text modeling.</p>
<p>CNN mainly contains convolution and pooling operation, with versatile combinations in various applications. Here, we briefly describe a CNN used to classify texts[<a href="#Refernce">1</a>], as shown in Figure 1.</p>
<p align="center">
<img src="/Users/xuxinlei02/Documents/forks/book/05.understand_sentiment/image/text_cnn_en.png" width = "80%" align="center"/><br/>
Figure 1. CNN for text modeling.
</p>

<p>Let $n$ be the length of the sentence to process, and the $i$-th word has embedding as $x_i\in\mathbb{R}^k$，where $k$ is the embedding dimensionality.</p>
<p>First, we concatenate the words by piecing together every $h$ words, each as a window of length $h$. This window is denoted as $x_{i:i+h-1}$, consisting of $x_{i},x_{i+1},\ldots,x_{i+h-1}$, where $x_i$ is the first word in the window and $i$ takes value ranging from $1$ to $n-h+1$: $x_{i:i+h-1}\in\mathbb{R}^{hk}$.</p>
<p>Next, we apply the convolution operation: we apply the kernel $w\in\mathbb{R}^{hk}$ in each window, extracting features $c_i=f(w\cdot x_{i:i+h-1}+b)$, where $b\in\mathbb{R}$ is the bias and $f$ is a non-linear activation function such as $sigmoid$. Convolving by the kernel at every window ${x_{1:h},x_{2:h+1},\ldots,x_{n-h+1:n}}$ produces a feature map in the following form:</p>
<p>$$c=[c_1,c_2,\ldots,c_{n-h+1}], c \in \mathbb{R}^{n-h+1}$$</p>
<p>Next, we apply <em>max pooling</em> over time to represent the whole sentence $\hat c$, which is the maximum element across the feature map:</p>
<p>$$\hat c=max(c)$$</p>
<p>In real applications, we will apply multiple CNN kernels on the sentences. It can be implemented efficiently by concatenating the kernels together as a matrix. Also, we can use CNN kernels with different kernel size (as shown in Figure 1 in different colors).</p>
<p>Finally, concatenating the resulting features produces a fixed-length representation, which can be combined with a softmax to form the model for the sentiment analysis problem.</p>
<p>For short texts, the aforementioned CNN model can achieve very high accuracy [<a href="#Reference">1</a>]. If we want to extract more abstract representations, we may apply a deeper CNN model [<a href="#Reference">2</a>,<a href="#Reference">3</a>].</p>
<h3 id="recurrent-neural-network-rnn"><a name="user-content-recurrent-neural-network-rnn" href="#recurrent-neural-network-rnn" class="headeranchor-link" aria-hidden="true"><span class="headeranchor"></span></a>Recurrent Neural Network (RNN)</h3>
<p>RNN is an effective model for sequential data. In terms of computability, the RNN is Turing-complete [<a href="#Reference">4</a>]. Since NLP is a classical problem on sequential data, the RNN, especially its variant LSTM[<a href="#Reference">5</a>]), achieves state-of-the-art performance on various NLP tasks, such as language modeling, syntax parsing, POS-tagging, image captioning, dialog, machine translation, and so forth.</p>
<p align="center">
<img src="/Users/xuxinlei02/Documents/forks/book/05.understand_sentiment/image/rnn.png" width = "60%" align="center"/><br/>
Figure 2. An illustration of an unfolded RNN in time.
</p>

<p>As shown in Figure 2, we unfold an RNN: at the $t$-th time step, the network takes two inputs: the $t$-th input vector $\vec{x_t}$ and the latent state from the last time-step $\vec{h_{t-1}}$. From those, it computes the latent state of the current step $\vec{h_t}$. This process is repeated until all inputs are consumed. Denoting the RNN as function $f$, it can be formulated as follows:</p>
<p>$$\vec{h_t}=f(\vec{x_t},\vec{h_{t-1}})=\sigma(W_{xh}\vec{x_t}+W_{hh}\vec{h_{h-1}}+\vec{b_h})$$</p>
<p>where $W_{xh}$ is the weight matrix to feed into the latent layer; $W_{hh}$ is the latent-to-latent matrix; $b_h$ is the latent bias and $\sigma$ refers to the $sigmoid$ function.</p>
<p>In NLP, words are often represented as a one-hot vectors and then mapped to an embedding. The embedded feature goes through an RNN as input $x_t$ at every time step. Moreover, we can add other layers on top of RNN, such as a deep or stacked RNN. Finally, the last latent state may be used as a feature for sentence classification.</p>
<h3 id="long-short-term-memory-lstm"><a name="user-content-long-short-term-memory-lstm" href="#long-short-term-memory-lstm" class="headeranchor-link" aria-hidden="true"><span class="headeranchor"></span></a>Long-Short Term Memory (LSTM)</h3>
<p>Training an RNN on long sequential data sometimes leads to the gradient vanishing or exploding[<a href="#">6</a>]. To solve this problem Hochreiter S, Schmidhuber J. (1997) proposed <strong>Long Short Term Memory</strong> (LSTM)[<a href="#Reference">5</a>]). </p>
<p>Compared to the structure of a simple RNN, an LSTM includes memory cell $c$, input gate $i$, forget gate $f$ and output gate $o$. These gates and memory cells dramatically improve the ability for the network to handle long sequences. We can formulate the <strong>LSTM-RNN</strong>, denoted as a function $F$, as follows：</p>
<p>$$ h_t=F(x_t,h_{t-1})$$</p>
<p>$F$ contains following formulations[<a href="#Reference">7</a>]：<br />
\begin{align}<br />
i_t &amp; = \sigma(W_{xi}x_t+W_{hi}h_{h-1}+W_{ci}c_{t-1}+b_i)\\<br />
f_t &amp; = \sigma(W_{xf}x_t+W_{hf}h_{h-1}+W_{cf}c_{t-1}+b_f)\\<br />
c_t &amp; = f_t\odot c_{t-1}+i_t\odot \tanh(W_{xc}x_t+W_{hc}h_{h-1}+b_c)\\<br />
o_t &amp; = \sigma(W_{xo}x_t+W_{ho}h_{h-1}+W_{co}c_{t}+b_o)\\<br />
h_t &amp; = o_t\odot \tanh(c_t)\\<br />
\end{align}</p>
<p>In the equation，$i_t, f_t, c_t, o_t$ stand for input gate, forget gate, memory cell and output gate, respectively; $W$ and $b$ are model parameters. The $tanh$ is a hyperbolic tangent， and $\odot$ denotes an element-wise product operation. Input gate controls the magnitude of new input into the memory cell $c$; forget gate controls memory propagated from the last time step; output gate controls output magnitude. The three gates are computed similarly with different parameters, and they influence memory cell $c$ separately, as shown in Figure 3:</p>
<p align="center">
<img src="/Users/xuxinlei02/Documents/forks/book/05.understand_sentiment/image/lstm_en.png" width = "65%" align="center"/><br/>
Figure 3. LSTM at time step $t$ [7].
</p>

<p>LSTM enhances the ability of considering long-term reliance, with the help of memory cell and gate. Similar structures are also proposed in Gated Recurrent Unit (GRU)[<a href="Reference">8</a>] with simpler design. <strong>The structures are still similar to RNN, though with some modifications (As shown in Figure 2), i.e., latent status depends on input as well as the latent status of last time-step, and the process goes on recurrently until all input are consumed:</strong></p>
<p>$$ h_t=Recrurent(x_t,h_{t-1})$$<br />
where $Recrurent$ is a simple RNN, GRU or LSTM.</p>
<h3 id="stacked-bidirectional-lstm"><a name="user-content-stacked-bidirectional-lstm" href="#stacked-bidirectional-lstm" class="headeranchor-link" aria-hidden="true"><span class="headeranchor"></span></a>Stacked Bidirectional LSTM</h3>
<p>For vanilla LSTM, $h_t$ contains input information from previous time-step $1..t-1$ context. We can also apply an RNN with reverse-direction to take successive context $t+1…n$ into consideration. Combining constructing deep RNN (deeper RNN can contain more abstract and higher level semantic), we can design structures with deep stacked bidirectional LSTM to model sequential data[<a href="#Reference">9</a>].</p>
<p>As shown in Figure 4 (3-layer RNN), odd/even layers are forward/reverse LSTM. Higher layers of LSTM take lower-layers LSTM as input, and the top-layer LSTM produces a fixed length vector by max-pooling (this representation considers contexts from previous and successive words for higher-level abstractions). Finally, we concatenate the output to a softmax layer for classification.</p>
<p align="center">
<img src="image/stacked_lstm_en.png" width=450><br/>
Figure 4. Stacked Bidirectional LSTM for NLP modeling.
</p>

<h2 id="dataset"><a name="user-content-dataset" href="#dataset" class="headeranchor-link" aria-hidden="true"><span class="headeranchor"></span></a>Dataset</h2>
<p>We use <a href="http://ai.stanford.edu/%7Eamaas/data/sentiment/">IMDB</a> dataset for sentiment analysis in this tutorial, which consists of 50,000 movie reviews split evenly into 25k train and 25k test sets. In the labeled train/test sets, a negative review has a score &lt;= 4 out of 10, and a positive review has a score &gt;= 7 out of 10.</p>
<p><code>paddle.datasets</code> package encapsulates multiple public datasets, including <code>cifar</code>, <code>imdb</code>, <code>mnist</code>, <code>moivelens</code>, and <code>wmt14</code>, etc. There&rsquo;s no need for us to manually download and preprocess IMDB.</p>
<p>After issuing a command <code>python train.py</code>, training will start immediately. The details will be unpacked by the following sessions to see how it works.</p>
<h2 id="model-structure"><a name="user-content-model-structure" href="#model-structure" class="headeranchor-link" aria-hidden="true"><span class="headeranchor"></span></a>Model Structure</h2>
<h3 id="initialize-paddlepaddle"><a name="user-content-initialize-paddlepaddle" href="#initialize-paddlepaddle" class="headeranchor-link" aria-hidden="true"><span class="headeranchor"></span></a>Initialize PaddlePaddle</h3>
<p>We must import and initialize PaddlePaddle (enable/disable GPU, set the number of trainers, etc).</p>
<pre><code class="python">import sys
import paddle.v2 as paddle

# PaddlePaddle init
paddle.init(use_gpu=False, trainer_count=1)
</code></pre>

<p>As alluded to in section <a href="#model-overview">Model Overview</a>, here we provide the implementations of both Text CNN and Stacked-bidirectional LSTM models.</p>
<h3 id="text-convolution-neural-network-text-cnn"><a name="user-content-text-convolution-neural-network-text-cnn" href="#text-convolution-neural-network-text-cnn" class="headeranchor-link" aria-hidden="true"><span class="headeranchor"></span></a>Text Convolution Neural Network (Text CNN)</h3>
<p>We create a neural network <code>convolution_net</code> as the following snippet code.</p>
<p>Note: <code>paddle.networks.sequence_conv_pool</code> includes both convolution and pooling layer operations.</p>
<pre><code class="python">def convolution_net(input_dim, class_dim=2, emb_dim=128, hid_dim=128):
    data = paddle.layer.data(&quot;word&quot;,
                             paddle.data_type.integer_value_sequence(input_dim))
    emb = paddle.layer.embedding(input=data, size=emb_dim)
    conv_3 = paddle.networks.sequence_conv_pool(
        input=emb, context_len=3, hidden_size=hid_dim)
    conv_4 = paddle.networks.sequence_conv_pool(
        input=emb, context_len=4, hidden_size=hid_dim)
    output = paddle.layer.fc(input=[conv_3, conv_4],
                             size=class_dim,
                             act=paddle.activation.Softmax())
    lbl = paddle.layer.data(&quot;label&quot;, paddle.data_type.integer_value(2))
    cost = paddle.layer.classification_cost(input=output, label=lbl)
    return cost
</code></pre>

<ol>
<li>
<p>Define input data and its dimension</p>
<p>Parameter <code>input_dim</code> denotes the dictionary size, and <code>class_dim</code> is the number of categories. In <code>convolution_net</code>, the input to the network is defined in <code>paddle.layer.data</code>.</p>
</li>
<li>
<p>Define Classifier</p>
<p>The above Text CNN network extracts high-level features and maps them to a vector of the same size as the categories. <code>paddle.activation.Softmax</code> function or classifier is then used for calculating the probability of the sentence belonging to each category.</p>
</li>
<li>
<p>Define Loss Function</p>
<p>In the context of supervised learning, labels of the training set are defined in <code>paddle.layer.data</code>, too. During training, cross-entropy is used as loss function in <code>paddle.layer.classification_cost</code> and as the output of the network; During testing, the outputs are the probabilities calculated in the classifier.</p>
</li>
</ol>
<h4 id="stacked-bidirectional-lstm_1"><a name="user-content-stacked-bidirectional-lstm_1" href="#stacked-bidirectional-lstm_1" class="headeranchor-link" aria-hidden="true"><span class="headeranchor"></span></a>Stacked bidirectional LSTM</h4>
<p>We create a neural network <code>stacked_lstm_net</code> as below.</p>
<pre><code class="python">def stacked_lstm_net(input_dim,
                     class_dim=2,
                     emb_dim=128,
                     hid_dim=512,
                     stacked_num=3):
    &quot;&quot;&quot;
    A Wrapper for sentiment classification task.
    This network uses bi-directional recurrent network,
    consisting three LSTM layers. This configure is referred to
    the paper as following url, but use fewer layrs.
        http://www.aclweb.org/anthology/P15-1109
    input_dim: here is word dictionary dimension.
    class_dim: number of categories.
    emb_dim: dimension of word embedding.
    hid_dim: dimension of hidden layer.
    stacked_num: number of stacked lstm-hidden layer.
    &quot;&quot;&quot;
    assert stacked_num % 2 == 1

    layer_attr = paddle.attr.Extra(drop_rate=0.5)
    fc_para_attr = paddle.attr.Param(learning_rate=1e-3)
    lstm_para_attr = paddle.attr.Param(initial_std=0., learning_rate=1.)
    para_attr = [fc_para_attr, lstm_para_attr]
    bias_attr = paddle.attr.Param(initial_std=0., l2_rate=0.)
    relu = paddle.activation.Relu()
    linear = paddle.activation.Linear()

    data = paddle.layer.data(&quot;word&quot;,
                             paddle.data_type.integer_value_sequence(input_dim))
    emb = paddle.layer.embedding(input=data, size=emb_dim)

    fc1 = paddle.layer.fc(input=emb,
                          size=hid_dim,
                          act=linear,
                          bias_attr=bias_attr)
    lstm1 = paddle.layer.lstmemory(
        input=fc1, act=relu, bias_attr=bias_attr, layer_attr=layer_attr)

    inputs = [fc1, lstm1]
    for i in range(2, stacked_num + 1):
        fc = paddle.layer.fc(input=inputs,
                             size=hid_dim,
                             act=linear,
                             param_attr=para_attr,
                             bias_attr=bias_attr)
        lstm = paddle.layer.lstmemory(
            input=fc,
            reverse=(i % 2) == 0,
            act=relu,
            bias_attr=bias_attr,
            layer_attr=layer_attr)
        inputs = [fc, lstm]

    fc_last = paddle.layer.pooling(
        input=inputs[0], pooling_type=paddle.pooling.Max())
    lstm_last = paddle.layer.pooling(
        input=inputs[1], pooling_type=paddle.pooling.Max())
    output = paddle.layer.fc(input=[fc_last, lstm_last],
                             size=class_dim,
                             act=paddle.activation.Softmax(),
                             bias_attr=bias_attr,
                             param_attr=para_attr)

    lbl = paddle.layer.data(&quot;label&quot;, paddle.data_type.integer_value(2))
    cost = paddle.layer.classification_cost(input=output, label=lbl)
    return cost
</code></pre>

<ol>
<li>
<p>Define input data and its dimension</p>
<p>Parameter <code>input_dim</code> denotes the dictionary size, and <code>class_dim</code> is the number of categories. In <code>stacked_lstm_net</code>, the input to the network is defined in <code>paddle.layer.data</code>.</p>
</li>
<li>
<p>Define Classifier</p>
<p>The above stacked bidirectional LSTM network extracts high-level features and maps them to a vector of the same size as the categories. <code>paddle.activation.Softmax</code> function or classifier is then used for calculating the probability of the sentence belonging to each category.</p>
</li>
<li>
<p>Define Loss Function</p>
<p>In the context of supervised learning, labels of the training set are defined in <code>paddle.layer.data</code>, too. During training, cross-entropy is used as loss function in <code>paddle.layer.classification_cost</code> and as the output of the network; During testing, the outputs are the probabilities calculated in the classifier.</p>
</li>
</ol>
<p>To reiterate, we can either invoke <code>convolution_net</code> or <code>stacked_lstm_net</code>.</p>
<pre><code class="python">word_dict = paddle.dataset.imdb.word_dict()
dict_dim = len(word_dict)
class_dim = 2

# option 1
cost = convolution_net(dict_dim, class_dim=class_dim)
# option 2
# cost = stacked_lstm_net(dict_dim, class_dim=class_dim, stacked_num=3)
</code></pre>

<h2 id="model-training"><a name="user-content-model-training" href="#model-training" class="headeranchor-link" aria-hidden="true"><span class="headeranchor"></span></a>Model Training</h2>
<h3 id="define-parameters"><a name="user-content-define-parameters" href="#define-parameters" class="headeranchor-link" aria-hidden="true"><span class="headeranchor"></span></a>Define Parameters</h3>
<p>First, we create the model parameters according to the previous model configuration <code>cost</code>.</p>
<pre><code class="python"># create parameters
parameters = paddle.parameters.create(cost)
</code></pre>

<h3 id="create-trainer"><a name="user-content-create-trainer" href="#create-trainer" class="headeranchor-link" aria-hidden="true"><span class="headeranchor"></span></a>Create Trainer</h3>
<p>Before jumping into creating a training module, algorithm setting is also necessary.<br />
Here we specified <code>Adam</code> optimization algorithm via <code>paddle.optimizer</code>.</p>
<pre><code class="python"># create optimizer
adam_optimizer = paddle.optimizer.Adam(
    learning_rate=2e-3,
    regularization=paddle.optimizer.L2Regularization(rate=8e-4),
    model_average=paddle.optimizer.ModelAverage(average_window=0.5))

# create trainer
trainer = paddle.trainer.SGD(cost=cost,
                                parameters=parameters,
                                update_equation=adam_optimizer)
</code></pre>

<h3 id="training"><a name="user-content-training" href="#training" class="headeranchor-link" aria-hidden="true"><span class="headeranchor"></span></a>Training</h3>
<p><code>paddle.dataset.imdb.train()</code> will yield records during each pass, after shuffling, a batch input is generated for training.</p>
<pre><code class="python">train_reader = paddle.batch(
    paddle.reader.shuffle(
        lambda: paddle.dataset.imdb.train(word_dict), buf_size=1000),
    batch_size=100)

test_reader = paddle.batch(
    lambda: paddle.dataset.imdb.test(word_dict), batch_size=100)
</code></pre>

<p><code>feeding</code> is devoted to specifying the correspondence between each yield record and <code>paddle.layer.data</code>. For instance, the first column of data generated by <code>paddle.dataset.imdb.train()</code> corresponds to <code>word</code> feature.</p>
<pre><code class="python">feeding = {'word': 0, 'label': 1}
</code></pre>

<p>Callback function <code>event_handler</code> will be invoked to track training progress when a pre-defined event happens.</p>
<pre><code class="python">def event_handler(event):
    if isinstance(event, paddle.event.EndIteration):
        if event.batch_id % 100 == 0:
            print &quot;\nPass %d, Batch %d, Cost %f, %s&quot; % (
                event.pass_id, event.batch_id, event.cost, event.metrics)
        else:
            sys.stdout.write('.')
            sys.stdout.flush()
    if isinstance(event, paddle.event.EndPass):
        result = trainer.test(reader=test_reader, feeding=feeding)
        print &quot;\nTest with Pass %d, %s&quot; % (event.pass_id, result.metrics)
</code></pre>

<p>Finally, we can invoke <code>trainer.train</code> to start training:</p>
<pre><code class="python">trainer.train(
    reader=train_reader,
    event_handler=event_handler,
    feeding=feeding,
    num_passes=10)
</code></pre>

<h2 id="conclusion"><a name="user-content-conclusion" href="#conclusion" class="headeranchor-link" aria-hidden="true"><span class="headeranchor"></span></a>Conclusion</h2>
<p>In this chapter, we use sentiment analysis as an example to introduce applying deep learning models on end-to-end short text classification, as well as how to use PaddlePaddle to implement the model. Meanwhile, we briefly introduce two models for text processing: CNN and RNN. In following chapters, we will see how these models can be applied in other tasks.</p>
<h2 id="reference"><a name="user-content-reference" href="#reference" class="headeranchor-link" aria-hidden="true"><span class="headeranchor"></span></a>Reference</h2>
<ol>
<li>Kim Y. <a href="http://arxiv.org/pdf/1408.5882">Convolutional neural networks for sentence classification</a>[J]. arXiv preprint arXiv:1408.5882, 2014.</li>
<li>Kalchbrenner N, Grefenstette E, Blunsom P. <a href="http://arxiv.org/pdf/1404.2188.pdf?utm_medium=App.net&amp;utm_source=PourOver">A convolutional neural network for modelling sentences</a>[J]. arXiv preprint arXiv:1404.2188, 2014.</li>
<li>Yann N. Dauphin, et al. <a href="https://arxiv.org/pdf/1612.08083v1.pdf">Language Modeling with Gated Convolutional Networks</a>[J] arXiv preprint arXiv:1612.08083, 2016.</li>
<li>Siegelmann H T, Sontag E D. <a href="http://research.cs.queensu.ca/home/akl/cisc879/papers/SELECTED_PAPERS_FROM_VARIOUS_SOURCES/05070215382317071.pdf">On the computational power of neural nets</a>[C]//Proceedings of the fifth annual workshop on Computational learning theory. ACM, 1992: 440-449.</li>
<li>Hochreiter S, Schmidhuber J. <a href="http://web.eecs.utk.edu/~itamar/courses/ECE-692/Bobby_paper1.pdf">Long short-term memory</a>[J]. Neural computation, 1997, 9(8): 1735-1780.</li>
<li>Bengio Y, Simard P, Frasconi P. <a href="http://www-dsi.ing.unifi.it/~paolo/ps/tnn-94-gradient.pdf">Learning long-term dependencies with gradient descent is difficult</a>[J]. IEEE transactions on neural networks, 1994, 5(2): 157-166.</li>
<li>Graves A. <a href="http://arxiv.org/pdf/1308.0850">Generating sequences with recurrent neural networks</a>[J]. arXiv preprint arXiv:1308.0850, 2013.</li>
<li>Cho K, Van Merriënboer B, Gulcehre C, et al. <a href="http://arxiv.org/pdf/1406.1078">Learning phrase representations using RNN encoder-decoder for statistical machine translation</a>[J]. arXiv preprint arXiv:1406.1078, 2014.</li>
<li>Zhou J, Xu W. <a href="http://www.aclweb.org/anthology/P/P15/P15-1109.pdf">End-to-end learning of semantic role labeling using recurrent neural networks</a>[C]//Proceedings of the Annual Meeting of the Association for Computational Linguistics. 2015.</li>
</ol>
<p><br/><br />
This tutorial is contributed by <a xmlns:cc="http://creativecommons.org/ns#" href="http://book.paddlepaddle.org" property="cc:attributionName" rel="cc:attributionURL">PaddlePaddle</a>, and licensed under a <a rel="license" href="http://creativecommons.org/licenses/by-nc-sa/4.0/">Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License</a>.</p></article></body></html>