Spaces:
Running
Running
| <html lang="en"> | |
| <head> | |
| <meta http-equiv="X-UA-Compatible" content="IE=Edge" /> | |
| <meta charset="utf-8" /> | |
| </head> | |
| <body style="margin: 0;"> | |
| <div id="p10" style="overflow: hidden; position: relative; background-color: white; width: 2200px; height: 1237px;"> | |
| <!-- Begin shared CSS values --> | |
| <style class="shared-css" type="text/css" > | |
| .t { | |
| transform-origin: bottom left; | |
| z-index: 2; | |
| position: absolute; | |
| white-space: pre; | |
| overflow: visible; | |
| line-height: 1.5; | |
| } | |
| .text-container { | |
| white-space: pre; | |
| } | |
| @supports (-webkit-touch-callout: none) { | |
| .text-container { | |
| white-space: normal; | |
| } | |
| } | |
| </style> | |
| <!-- End shared CSS values --> | |
| <!-- Begin inline CSS --> | |
| <style type="text/css" > | |
| #t1_10{left:0px;bottom:1054px;letter-spacing:-0.23px;} | |
| #t2_10{left:0px;bottom:917px;letter-spacing:-0.24px;} | |
| #t3_10{left:0px;bottom:781px;letter-spacing:-0.24px;} | |
| #t4_10{left:0px;bottom:645px;letter-spacing:-0.21px;} | |
| #t5_10{left:565px;bottom:998px;letter-spacing:0.21px;} | |
| #t6_10{left:565px;bottom:934px;letter-spacing:0.2px;} | |
| #t7_10{left:565px;bottom:870px;letter-spacing:0.2px;} | |
| #t8_10{left:699px;bottom:1080px;letter-spacing:0.14px;} | |
| #t9_10{left:565px;bottom:766px;letter-spacing:0.21px;} | |
| #ta_10{left:861px;bottom:766px;letter-spacing:0.22px;} | |
| #tb_10{left:565px;bottom:702px;letter-spacing:0.2px;} | |
| #tc_10{left:101px;bottom:501px;letter-spacing:0.21px;} | |
| #td_10{left:398px;bottom:501px;letter-spacing:0.2px;} | |
| #te_10{left:101px;bottom:437px;letter-spacing:0.2px;} | |
| #tf_10{left:101px;bottom:373px;letter-spacing:0.21px;} | |
| #tg_10{left:101px;bottom:309px;letter-spacing:0.2px;} | |
| #th_10{left:101px;bottom:245px;letter-spacing:0.21px;} | |
| #ti_10{left:101px;bottom:181px;letter-spacing:0.21px;} | |
| .s1_10{font-size:115px;font-family:IBMPlexSans_2d;color:#000;} | |
| .s2_10{font-size:115px;font-family:IBMPlexSans-Bold_2l;color:#000;} | |
| .s3_10{font-size:53px;font-family:IBMPlexSans-Italic_2o;color:#000;} | |
| .s4_10{font-size:99px;font-family:IBMPlexSans-Bold_2l;color:#000;} | |
| .s5_10{font-size:53px;font-family:IBMPlexSans_2d;color:#000;} | |
| </style> | |
| <!-- End inline CSS --> | |
| <!-- Begin embedded font definitions --> | |
| <style id="fonts10" type="text/css" > | |
| @font-face { | |
| font-family: IBMPlexSans-Bold_2l; | |
| src: url("fonts/IBMPlexSans-Bold_2l.woff") format("woff"); | |
| } | |
| @font-face { | |
| font-family: IBMPlexSans-Italic_2o; | |
| src: url("fonts/IBMPlexSans-Italic_2o.woff") format("woff"); | |
| } | |
| @font-face { | |
| font-family: IBMPlexSans_2d; | |
| src: url("fonts/IBMPlexSans_2d.woff") format("woff"); | |
| } | |
| </style> | |
| <!-- End embedded font definitions --> | |
| <!-- Begin page background --> | |
| <div id="pg10Overlay" style="width:100%; height:100%; position:absolute; z-index:1; background-color:rgba(0,0,0,0); -webkit-user-select: none;"></div> | |
| <div id="pg10" style="-webkit-user-select: none;"><object width="2200" height="1237" data="10/10.svg" type="image/svg+xml" id="pdf10" style="width:2200px; height:1237px; -moz-transform:scale(1); z-index: 0;"></object></div> | |
| <!-- End page background --> | |
| <!-- Begin text definitions (Positioned/styled in CSS) --> | |
| <div class="text-container"><span id="t1_10" class="t s1_10">Problems </span> | |
| <span id="t2_10" class="t s1_10">and </span> | |
| <span id="t3_10" class="t s2_10">Proposed </span> | |
| <span id="t4_10" class="t s2_10">Solutions </span> | |
| <span id="t5_10" class="t s3_10">When a model has learned to focus too much on the specific </span> | |
| <span id="t6_10" class="t s3_10">characteristics of the training data, rather than generalizing to new </span> | |
| <span id="t7_10" class="t s3_10">situations. </span> | |
| <span id="t8_10" class="t s4_10">2.Overfitting </span> | |
| <span id="t9_10" class="t s5_10">Solution 1 </span><span id="ta_10" class="t s5_10">Dreambooth used prior-preservation loss, and the </span> | |
| <span id="tb_10" class="t s5_10">ratio of prior-preservation is never easy to determine. </span> | |
| <span id="tc_10" class="t s5_10">Solution 2 </span><span id="td_10" class="t s5_10">It is training data that caused the overfitting. Thus we use a subset of </span> | |
| <span id="te_10" class="t s5_10">the training data to train an overfitted model, select the previous checkpoint which </span> | |
| <span id="tf_10" class="t s5_10">and use it to generate images by prompt for a single word. These images can be </span> | |
| <span id="tg_10" class="t s5_10">placed in the regular training data according to the word frequency ratio, and the </span> | |
| <span id="th_10" class="t s5_10">subset of the which that caused the overfitting can be removed, and then retrain </span> | |
| <span id="ti_10" class="t s5_10">the model from the very beginning. </span></div> | |
| <!-- End text definitions --> | |
| </div> | |
| </body> | |
| </html> | |