Add custom tokenizers

src/lang/en.ts

@@ -576,4 +576,5 @@ export const languageEnglish = {
     autoContinueMinTokens: "Target Tokens (Auto Continue)",
     autoContinueChat: "Anti-Incomplete Response (Auto Continue)",
     removeIncompleteResponse: "Remove Incomplete Sentences",
+    tokenizer: "Tokenizer",
 }

src/lib/Setting/Pages/BotSettings.svelte

@@ -5,7 +5,7 @@
     import { DataBase } from "src/ts/storage/database";
     import { customProviderStore, getCurrentPluginMax } from "src/ts/plugins/plugins";
     import { getModelMaxContext, isTauri } from "src/ts/storage/globalApi";
-    import { tokenize, tokenizeAccurate } from "src/ts/tokenizer";
+    import { tokenize, tokenizeAccurate, tokenizerList } from "src/ts/tokenizer";
     import ModelList from "src/lib/UI/ModelList.svelte";
     import DropList from "src/lib/SideBars/DropList.svelte";
     import { PlusIcon, TrashIcon } from "lucide-svelte";
@@ -242,6 +242,14 @@
         </SelectInput>
     {/await}
 {/if}
+{#if $DataBase.aiModel === 'openrouter' || $DataBase.aiModel === 'reverse_proxy'}
+    <span class="text-textcolor">{language.tokenizer}</span>
+    <SelectInput bind:value={$DataBase.customTokenizer}>
+        {#each tokenizerList as entry}
+            <OptionInput value={entry[0]}>{entry[1]}</OptionInput>
+        {/each}
+    </SelectInput>
+{/if}
 {#if $DataBase.aiModel.startsWith('gpt') || $DataBase.subModel.startsWith('gpt')
     || $DataBase.aiModel.startsWith('instructgpt') || $DataBase.subModel.startsWith('instructgpt')}
     <span class="text-textcolor">OpenAI {language.apiKey} <Help key="oaiapikey"/></span>
@@ -254,10 +262,10 @@
     </div>
 {/if}
 {#if $DataBase.aiModel.startsWith('openrouter')}
-    <div class="flex items-center">
+    <div class="flex items-center mb-4">
         <Check bind:check={$DataBase.openrouterFallback} name={language.openrouterFallback}/>
     </div>
-    <div class="flex items-center">
+    <div class="flex items-center mb-4">
         <Check bind:check={$DataBase.openrouterMiddleOut} name={language.openrouterMiddleOut}/>
     </div>
 {/if}

src/ts/storage/database.ts

@@ -390,6 +390,7 @@ export function setDatabase(data:Database){
     data.repetition_penalty ??= 1
     data.min_p ??= 0
     data.top_a ??= 0
+    data.customTokenizer ??= 'tik'
 
     changeLanguage(data.language)
     DataBase.set(data)
@@ -633,6 +634,7 @@ export interface Database{
     autoContinueChat:boolean
     autoContinueMinTokens:number
     removeIncompleteResponse:boolean
+    customTokenizer:string
 }
 
 export interface customscript{

src/ts/tokenizer.ts

@@ -7,8 +7,34 @@ import { supportsInlayImage } from "./process/files/image";
 import { risuChatParser } from "./parser";
 import { tokenizeGGUFModel } from "./process/models/local";
 
+
+export const tokenizerList = [
+    ['tik', 'Tiktoken (OpenAI)'],
+    ['mistral', 'Mistral'],
+    ['novelai', 'NovelAI'],
+    ['claude', 'Claude'],
+    ['llama', 'Llama'],
+    ['novellist', 'Novellist'],
+] as const
+
 async function encode(data:string):Promise<(number[]|Uint32Array|Int32Array)>{
     let db = get(DataBase)
+    if(db.aiModel === 'openrouter' || db.aiModel === 'reverse_proxy'){
+        switch(db.customTokenizer){
+            case 'mistral':
+                return await tokenizeWebTokenizers(data, 'mistral')
+            case 'llama':
+                return await tokenizeWebTokenizers(data, 'llama')
+            case 'novelai':
+                return await tokenizeWebTokenizers(data, 'novelai')
+            case 'claude':
+                return await tokenizeWebTokenizers(data, 'claude')
+            case 'novellist':
+                return await tokenizeWebTokenizers(data, 'novellist')
+            default:
+                return await tikJS(data)
+        }
+    }
     if(db.aiModel.startsWith('novellist')){
         const nv= await tokenizeWebTokenizers(data, 'novellist')
         return nv