| 1471 | 1471 | <div id="banned_tokens_block_ooba" data-newbie-hidden class="wide100p"> |
| 1472 | 1472 | <hr data-newbie-hidden class="width100p"> |
| 1473 | 1473 | <h4 class="range-block-title justifyCenter"> |
| 1474 | 1474 | <span data-i18n="Banned Tokens">Banned Tokens/Strings</span> |
| 1475 | 1475 | <div class="margin5 fa-solid fa-circle-info opacity50p " data-i18n="[title]LLaMA / Mistral / Yi models only" title="LLaMA / Mistral / Yi models only. Make sure to select an appropriate tokenizerEnter first. Sequencessequences you don't want to appear in the output. OneUnquoted pertext line.will Textbe ortokenized in the back end and banned as tokens. [token ids] will be banned as-is. Most tokens have a leading space. Use token counter (with the correct tokenizer selected first!) if you are unsure. Enclose text in double quotes to ban the entire string as a set. Quoted Strings and [Token ids] must be on their own line."></div> |
| 1476 | 1476 | </h4> |
| 1477 | 1477 | <div class="wide100p"> |
| 1478 | 1478 | <textarea id="banned_tokens_textgenerationwebui" class="text_pole textarea_compact" name="banned_tokens_textgenerationwebui" rows="3" data-i18n="[placeholder]Example: some text [42, 69, 1337]" placeholder="Example: 'some text as tokens [42420, 69, 1337] "Some verbatim string"'></textarea> |
| 1479 | 1479 | </div> |
| 1480 | 1480 | </div> |
| 1481 | 1481 | <div class="range-block wide100p"> |