Add proper LLaMA tokenizer and tokenizer switch. Remove legacy code.

This commit is contained in:
SillyLossy
2023-04-26 00:09:45 +03:00
parent 655885b1a2
commit 2ae28023c0
6 changed files with 77 additions and 193 deletions

View File

@ -1037,6 +1037,15 @@
</div>
<div name="ContextFormatting">
<h4>Context Formatting</h4>
<div>
<h4>Tokenizer</h4>
<select id="tokenizer">
<option value="0">None / Estimated</option>
<option value="1">GPT-3 (OpenAI)</option>
<option value="2">GPT-3 (Alternative / Classic)</option>
<option value="3">Sentencepiece (LLaMA)</option>
</select>
</div>
<label class="checkbox_label" for="always-force-name2-checkbox">
<input id="always-force-name2-checkbox" type="checkbox" />
Always add character's name to prompt

View File

@ -48,6 +48,7 @@ import {
sortCharactersList,
power_user,
pygmalion_options,
tokenizers,
} from "./scripts/power-user.js";
import {
@ -335,14 +336,31 @@ $(document).ajaxError(function myErrorHandler(_, xhr) {
});
function getTokenCount(str, padding = 0) {
if (main_api == 'poe' || main_api == 'openai') {
switch (power_user.tokenizer) {
case tokenizers.NONE:
return Math.ceil(str.length / CHARACTERS_PER_TOKEN_RATIO) + padding;
case tokenizers.GPT3:
return gpt3.encode(str).bpe.length + padding;
}
else {
case tokenizers.CLASSIC:
return encode(str).length + padding;
case tokenizers.LLAMA:
let tokenCount = 0;
jQuery.ajax({
async: false,
type: 'POST', //
url: `/tokenize_llama`,
data: JSON.stringify({ text: str }),
dataType: "json",
contentType: "application/json",
success: function (data) {
tokenCount = data.count;
}
});
return tokenCount + padding;
}
}
const CHARACTERS_PER_TOKEN_RATIO = 3.35;
const talkativeness_default = 0.5;
var is_advanced_char_open = false;

View File

@ -6,8 +6,6 @@ import {
token,
getStatus,
} from "../script.js";
import { delay } from "./utils.js";
export {
loadPowerUserSettings,
@ -16,6 +14,7 @@ export {
sortCharactersList,
power_user,
pygmalion_options,
tokenizers,
};
const avatar_styles = {
@ -39,7 +38,15 @@ const pygmalion_options = {
ENABLED: 1,
}
const tokenizers = {
NONE: 0,
GPT3: 1,
CLASSIC: 2,
LLAMA: 3,
}
let power_user = {
tokenizer: tokenizers.CLASSIC,
collapse_newlines: false,
pygmalion_formatting: pygmalion_options.AUTO,
pin_examples: false,
@ -97,7 +104,6 @@ const storage_keys = {
movingUI: "TavernAI_movingUI",
};
const chat = document.getElementById('chat');
let browser_has_focus = true;
function playMessageSound() {
@ -260,6 +266,7 @@ function loadPowerUserSettings(settings, data) {
power_user.font_scale = Number(localStorage.getItem(storage_keys.font_scale) ?? 1);
power_user.blur_strength = Number(localStorage.getItem(storage_keys.blur_strength) ?? 10);
$(`#tokenizer option[value="${power_user.tokenizer}"]`).attr('selected', true);
$(`#pygmalion_formatting option[value=${power_user.pygmalion_formatting}]`).attr("selected", true);
$("#collapse-newlines-checkbox").prop("checked", power_user.collapse_newlines);
$("#pin-examples-checkbox").prop("checked", power_user.pin_examples);
@ -584,6 +591,12 @@ $(document).ready(() => {
saveSettingsDebounced();
});
$("#tokenizer").on('change', function () {
const value = $(this).find(':selected').val();
power_user.tokenizer = Number(value);
saveSettingsDebounced();
});
$(window).on('focus', function () {
browser_has_focus = true;
});

206
server.js
View File

@ -119,6 +119,16 @@ let api_key_openai;
const delay = ms => new Promise(resolve => setTimeout(resolve, ms))
const { SentencePieceProcessor, cleanText } = require("./src/sentencepiece/sentencepiece.min.js");
let spp = new SentencePieceProcessor();
async function countTokensLlama(text) {
let cleaned = cleanText(text);
let ids = spp.encodeIds(cleaned);
return ids.length;
}
function humanizedISO8601DateTime() {
let baseDate = new Date(Date.now());
let humanYear = baseDate.getFullYear();
@ -2273,6 +2283,15 @@ app.post("/savepreset_openai", jsonParser, function (request, response) {
return response.send({ name });
});
app.post("/tokenize_llama", jsonParser, async function (request, response) {
if (!request.body) {
return response.sendStatus(400);
}
const count = await countTokensLlama(request.body.text);
return response.send({ count });
});
// ** REST CLIENT ASYNC WRAPPERS **
function deleteAsync(url, args) {
return new Promise((resolve, reject) => {
@ -2338,6 +2357,8 @@ const setupTasks = async function () {
// Colab users could run the embedded tool
if (!is_colab) await convertWebp();
await spp.load(`./src/sentencepiece/tokenizer.model`);
console.log('Launching...');
if (autorun) open(autorunUrl.toString());
@ -2346,10 +2367,6 @@ const setupTasks = async function () {
!config.whitelistMode &&
!config.basicAuthMode)
console.log('Your SillyTavern is currently open to the public. To increase security, consider enabling whitelisting or basic authentication.')
if (fs.existsSync('public/characters/update.txt') && !is_colab) {
convertStage1();
}
}
if (true === cliArguments.ssl)
@ -2370,187 +2387,6 @@ else
setupTasks
);
//#####################CONVERTING IN NEW FORMAT########################
var charactersB = {};//B - Backup
var character_ib = 0;
var directoriesB = {};
function convertStage1() {
//if (!fs.existsSync('public/charactersBackup')) {
//fs.mkdirSync('public/charactersBackup');
//copyFolder('public/characters/', 'public/charactersBackup');
//}
var directories = getDirectories2("public/characters");
//console.log(directories[0]);
charactersB = {};
character_ib = 0;
var folderForDel = {};
getCharacterFile2(directories, 0);
}
function convertStage2() {
var mes = true;
for (const key in directoriesB) {
if (mes) {
console.log('***');
console.log('The update of the character format has begun...');
console.log('***');
mes = false;
}
var char = JSON.parse(charactersB[key]);
char.create_date = humanizedISO8601DateTime();
charactersB[key] = JSON.stringify(char);
var avatar = 'public/img/ai4.png';
if (char.avatar !== 'none') {
avatar = 'public/characters/' + char.name + '/avatars/' + char.avatar;
}
charaWrite(avatar, charactersB[key], directoriesB[key]);
const files = fs.readdirSync('public/characters/' + directoriesB[key] + '/chats');
if (!fs.existsSync(chatsPath + char.name)) {
fs.mkdirSync(chatsPath + char.name);
}
files.forEach(function (file) {
// Read the contents of the file
const fileContents = fs.readFileSync('public/characters/' + directoriesB[key] + '/chats/' + file, 'utf8');
// Iterate through the array of strings and parse each line as JSON
let chat_data = JSON.parse(fileContents);
let new_chat_data = [];
let this_chat_user_name = 'You';
let is_pass_0 = false;
if (chat_data[0].indexOf('<username-holder>') !== -1) {
this_chat_user_name = chat_data[0].substr('<username-holder>'.length, chat_data[0].length);
is_pass_0 = true;
}
let i = 0;
let ii = 0;
new_chat_data[i] = { user_name: 'You', character_name: char.name, create_date: humanizedISO8601DateTime() };
i++;
ii++;
chat_data.forEach(function (mes) {
if (!(i === 1 && is_pass_0)) {
if (mes.indexOf('<username-holder>') === -1 && mes.indexOf('<username-idkey>') === -1) {
new_chat_data[ii] = {};
let is_name = false;
if (mes.trim().indexOf(this_chat_user_name + ':') !== 0) {
if (mes.trim().indexOf(char.name + ':') === 0) {
mes = mes.replace(char.name + ':', '');
is_name = true;
}
new_chat_data[ii]['name'] = char.name;
new_chat_data[ii]['is_user'] = false;
new_chat_data[ii]['is_name'] = is_name;
new_chat_data[ii]['send_date'] = humanizedISO8601DateTime(); //Date.now();
} else {
mes = mes.replace(this_chat_user_name + ':', '');
new_chat_data[ii]['name'] = 'You';
new_chat_data[ii]['is_user'] = true;
new_chat_data[ii]['is_name'] = true;
new_chat_data[ii]['send_date'] = humanizedISO8601DateTime(); //Date.now();
}
new_chat_data[ii]['mes'] = mes.trim();
ii++;
}
}
i++;
});
const jsonlData = new_chat_data.map(JSON.stringify).join('\n');
// Write the contents to the destination folder
//console.log('convertstage2 writing a file: '+chatsPath+char.name+'/' + file+'l');
fs.writeFileSync(chatsPath + char.name + '/' + file + 'l', jsonlData);
});
//fs.rmSync('public/characters/'+directoriesB[key],{ recursive: true });
console.log(char.name + ' update!');
}
//removeFolders('public/characters');
fs.unlinkSync('public/characters/update.txt');
if (mes == false) {
console.log('***');
console.log('Сharacter format update completed successfully!');
console.log('***');
console.log('Now you can delete these folders, they are no longer used by TavernAI:');
}
for (const key in directoriesB) {
console.log('public/characters/' + directoriesB[key]);
}
}
function removeFolders(folder) {
const files = fs.readdirSync(folder);
files.forEach(function (file) {
const filePath = folder + '/' + file;
const stat = fs.statSync(filePath);
if (stat.isDirectory()) {
removeFolders(filePath);
fs.rmdirSync(filePath);
}
});
}
function copyFolder(src, dest) {
const files = fs.readdirSync(src);
files.forEach(function (file) {
const filePath = src + '/' + file;
const stat = fs.statSync(filePath);
if (stat.isFile()) {
fs.copyFileSync(filePath, dest + '/' + file);
} else if (stat.isDirectory()) {
fs.mkdirSync(dest + '/' + file);
copyFolder(filePath, dest + '/' + file);
}
});
}
function getDirectories2(path) {
return fs.readdirSync(path)
.filter(function (file) {
return fs.statSync(path + '/' + file).isDirectory();
})
.sort(function (a, b) {
return new Date(fs.statSync(path + '/' + a).mtime) - new Date(fs.statSync(path + '/' + b).mtime);
})
.reverse();
}
function getCharacterFile2(directories, i) {
if (directories.length > i) {
fs.stat('public/characters/' + directories[i] + '/' + directories[i] + ".json", function (err, stat) {
if (err == null) {
fs.readFile('public/characters/' + directories[i] + '/' + directories[i] + ".json", 'utf8', (err, data) => {
if (err) {
console.error(err);
return;
}
//console.log(data);
if (!fs.existsSync('public/characters/' + directories[i] + '.png')) {
charactersB[character_ib] = {};
charactersB[character_ib] = data;
directoriesB[character_ib] = directories[i];
character_ib++;
}
i++;
getCharacterFile2(directories, i);
});
} else {
i++;
getCharacterFile2(directories, i);
}
});
} else {
convertStage2();
}
}
async function convertWebp() {
const files = fs.readdirSync(directories.characters).filter(e => e.endsWith(".webp"));

File diff suppressed because one or more lines are too long

Binary file not shown.