feat: Add max tokens setting for model generations

Adds a new setting to control the maximum number of tokens generated by the model. This provides more control over the length of responses and can be useful for limiting the amount of text generated in certain situations.

src/assets/locale/da/common.json

@@ -70,6 +70,10 @@
                 "label": "Længden af Kontekst",
                 "placeholder": "Instast Længden af Kontekst værdien (standard: 2048)"
             },
+            "numPredict": {
+                "label": "Maks Tokens (num_predict)",
+                "placeholder": "Indtast Maks Tokens værdi (fx. 2048, 4096)"
+            },
             "seed": {
                 "label": "Seed",
                 "placeholder": "Indtast Seed værdi (fx. 1234)",

src/assets/locale/de/common.json

@@ -70,6 +70,10 @@
                 "label": "Anzahl der Kontexte",
                 "placeholder": "Geben Sie die Anzahl der Kontexte ein (Standard: 2048)"
             },
+            "numPredict": {
+                "label": "Max Tokens (num_predict)",
+                "placeholder": "Geben Sie den Max-Tokens-Wert ein (z.B. 2048, 4096)"
+            },
             "seed": {
                 "label": "Seed",
                 "placeholder": "Geben Sie den Seed-Wert ein (z.B. 1234)",

src/assets/locale/en/common.json

@@ -70,6 +70,10 @@
                 "label": "Number of Contexts",
                 "placeholder": "Enter Number of Contexts value (default: 2048)"
             },
+            "numPredict": {
+                "label": "Max Tokens (num_predict)",
+                "placeholder": "Enter Max Tokens value (e.g. 2048, 4096)"
+            },
             "seed": {
                 "label": "Seed",
                 "placeholder": "Enter Seed value (e.g. 1234)",

src/assets/locale/es/common.json

@@ -70,6 +70,10 @@
                 "label": "Cantidad de contextos",
                 "placeholder": "Ingresar el valor de tamaño de la ventana de contexto (por defecto: 2048)"
             },
+            "numPredict": {
+                "label": "Máximo de Tokens (num_predict)",
+                "placeholder": "Ingrese el valor máximo de Tokens (ej: 2048, 4096)"
+            },
             "seed": {
                 "label": "Semilla",
                 "placeholder": "Ingresar el valor de la semilla (ej: 1234)",

src/assets/locale/fa/common.json

@@ -70,6 +70,10 @@
                 "label": "Number of Contexts",
                 "placeholder": "مقدار Number of Contexts را وارد کنید (پیش فرض: 2048)"
             },
+            "numPredict": {
+                "label": "حداکثر توکن‌ها (num_predict)",
+                "placeholder": "مقدار حداکثر توکن‌ها را وارد کنید (مثلا 2048، 4096)"
+            },
             "seed": {
                 "label": "Seed",
                 "placeholder": "مقدار Seed را وارد کنید (e.g. 1234)",

src/assets/locale/fr/common.json

@@ -70,6 +70,10 @@
                 "label": "Nombre de contextes",
                 "placeholder": "Entrez la valeur du nombre de contextes (par défaut: 2048)"
             },
+            "numPredict": {
+                "label": "Tokens maximum (num_predict)",
+                "placeholder": "Entrez la valeur des tokens maximum (par exemple 2048, 4096)"
+            },
             "seed": {
                 "label": "Graine",
                 "placeholder": "Entrez la valeur des semences (par exemple 1234)",

src/assets/locale/it/common.json

@@ -70,6 +70,10 @@
                 "label": "Dimensione del Contesto",
                 "placeholder": "Inserisci la Dimensione del Contesto (default: 2048)"
             },
+            "numPredict": {
+                "label": "Token Massimi (num_predict)",
+                "placeholder": "Inserisci il valore dei Token Massimi (es. 2048, 4096)"
+            },
             "seed": {
                 "label": "Seed",
                 "placeholder": "Inserisci il Valore Seed (e.g. 1234)",

src/assets/locale/ja-JP/common.json

@@ -70,6 +70,10 @@
         "label": "コンテキストの数",
         "placeholder": "コンテキスト数を入力してください（デフォルト：2048）"
       },
+      "numPredict": {
+        "label": "最大トークン数 (num_predict)",
+        "placeholder": "最大トークン数を入力してください（例：2048、4096）"
+      },
       "seed": {
         "label": "シード",
         "placeholder": "シード値を入力してください（例：1234）",

src/assets/locale/ko/common.json

@@ -70,7 +70,10 @@
         "label": "컨텍스트 수",
         "placeholder": "컨텍스트 수를 입력하세요 (기본값: 2048)"
       },
-      "seed": {
+      "numPredict": {
+        "label": "최대 토큰 수 (num_predict)",
+        "placeholder": "최대 토큰 수를 입력하세요 (예: 2048, 4096)"
+      },      "seed": {
         "label": "시드",
         "placeholder": "시드 값을 입력하세요 (예: 1234)",
         "help": "모델 출력의 재현성"

src/assets/locale/ml/common.json

@@ -69,6 +69,10 @@
                 "label": "സന്ദർഭങ്ങളുടെ എണ്ണം",
                 "placeholder": "സന്ദർഭങ്ങളുടെ സംഖ്യ നൽകുക (സ്ഥിരം: 2048)"
             },
+            "numPredict": {
+                "label": "പരമാവധി ടോക്കണുകൾ (num_predict)",
+                "placeholder": "പരമാവധി ടോക്കൺ മൂല്യം നൽകുക (ഉദാ: 2048, 4096)"
+            },
             "seed": {
                 "label": "സീഡ്",
                 "placeholder": "സീഡ് വില്യമ നൽകുക (ഉദാ: 1234)",

src/assets/locale/no/common.json

@@ -70,6 +70,10 @@
                 "label": "Kontekstlengde",
                 "placeholder": "Skriv inn kontekstlengdeverdi (standard: 2048)"
             },
+            "numPredict": {
+                "label": "Maks Tokens (num_predict)",
+                "placeholder": "Skriv inn Maks Tokens-verdi (f.eks. 2048, 4096)"
+            },
             "seed": {
                 "label": "Seed",
                 "placeholder": "Skriv inn seedverdi (f.eks. 1234)",

src/assets/locale/pt-BR/common.json

@@ -70,6 +70,10 @@
         "label": "Número de Contextos",
         "placeholder": "Digite o valor do Número de Contextos (padrão: 2048)"
       },
+      "numPredict": {
+        "label": "Máximo de Tokens (num_predict)",
+        "placeholder": "Digite o valor do Máximo de Tokens (ex: 2048, 4096)"
+      },
       "seed": {
         "label": "Semente",
         "placeholder": "Digite o valor da Semente (ex: 1234)",

src/assets/locale/ru/common.json

@@ -70,6 +70,10 @@
                 "label": "Количество контекстов",
                 "placeholder": "Введите значение количества контекстов (по умолчанию: 2048)"
             },
+            "numPredict": {
+                "label": "Максимальное количество токенов (num_predict)",
+                "placeholder": "Введите значение максимального количества токенов (например, 2048, 4096)"
+            },
             "seed": {
                 "label": "Сид",
                 "placeholder": "Введите значение сида (например, 1234)",

src/assets/locale/sv/common.json

@@ -70,6 +70,10 @@
                 "label": "Antal kontexter",
                 "placeholder": "Ange antal kontextvärden (standard: 2048)"
             },
+            "numPredict": {
+                "label": "Max antal tokens (num_predict)",
+                "placeholder": "Ange Max antal tokens värde (t.ex. 2048, 4096)"
+            },
             "seed": {
                 "label": "Frö",
                 "placeholder": "Ange frövärde (t.ex. 1234)",

src/assets/locale/zh/common.json

@@ -70,6 +70,10 @@
                 "label": "上下文数量",
                 "placeholder": "输入上下文数量（默认：2048）"
             },
+            "numPredict": {
+                "label": "最大令牌数 (num_predict)",
+                "placeholder": "输入最大令牌数（例如：2048、4096）"
+            },
             "seed": {
                 "label": "随机种子",
                 "placeholder": "输入随机种子值（例如：1234）",

src/components/Common/Settings/CurrentChatModelSettings.tsx

@@ -39,12 +39,14 @@ export const CurrentChatModelSettings = ({
         numCtx: cUserSettings.numCtx ?? data.numCtx,
         seed: cUserSettings.seed,
         numGpu: cUserSettings.numGpu ?? data.numGpu,
+        numPredict: cUserSettings.numPredict ?? data.numPredict,
         systemPrompt: cUserSettings.systemPrompt ?? ""
       })
       return data
     },
     enabled: open,
-    refetchOnMount: true
+    refetchOnMount: false,
+    refetchOnWindowFocus: false
   })
 
   const renderBody = () => {
@@ -115,6 +117,15 @@ export const CurrentChatModelSettings = ({
               />
             </Form.Item>
 
+            <Form.Item
+              name="numPredict"
+              label={t("modelSettings.form.numPredict.label")}>
+              <InputNumber
+                style={{ width: "100%" }}
+                placeholder={t("modelSettings.form.numPredict.placeholder")}
+              />
+            </Form.Item>
+
             <Collapse
               ghost
               className="border-none bg-transparent"

src/components/Option/Settings/model-settings.tsx

@@ -77,7 +77,14 @@ export const ModelSettings = () => {
               size="large"
             />
           </Form.Item>
-
+          <Form.Item
+              name="numPredict"
+              label={t("modelSettings.form.numPredict.label")}>
+              <InputNumber
+                style={{ width: "100%" }}
+                placeholder={t("modelSettings.form.numPredict.placeholder")}
+              />
+            </Form.Item>
           <Collapse
             ghost
             className="border-none bg-transparent"

src/hooks/useMessage.tsx

@@ -133,7 +133,9 @@ export const useMessage = () => {
         currentChatModelSettings?.numCtx ?? userDefaultModelSettings?.numCtx,
       seed: currentChatModelSettings?.seed,
       numGpu:
-        currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu
+        currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu,
+       numPredict: currentChatModelSettings?.numPredict ?? userDefaultModelSettings?.numPredict,
+
     })
 
     let newMessage: Message[] = []
@@ -261,7 +263,9 @@ export const useMessage = () => {
             userDefaultModelSettings?.numCtx,
           seed: currentChatModelSettings?.seed,
           numGpu:
-            currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu
+            currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu,
+       numPredict: currentChatModelSettings?.numPredict ?? userDefaultModelSettings?.numPredict,
+
         })
         const response = await questionOllama.invoke(promptForQuestion)
         query = response.content.toString()
@@ -475,7 +479,9 @@ export const useMessage = () => {
         currentChatModelSettings?.numCtx ?? userDefaultModelSettings?.numCtx,
       seed: currentChatModelSettings?.seed,
       numGpu:
-        currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu
+        currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu,
+       numPredict: currentChatModelSettings?.numPredict ?? userDefaultModelSettings?.numPredict,
+
     })
 
     let newMessage: Message[] = []
@@ -702,7 +708,9 @@ export const useMessage = () => {
         currentChatModelSettings?.numCtx ?? userDefaultModelSettings?.numCtx,
       seed: currentChatModelSettings?.seed,
       numGpu:
-        currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu
+        currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu,
+       numPredict: currentChatModelSettings?.numPredict ?? userDefaultModelSettings?.numPredict,
+
     })
 
     let newMessage: Message[] = []
@@ -777,7 +785,9 @@ export const useMessage = () => {
             userDefaultModelSettings?.numCtx,
           seed: currentChatModelSettings?.seed,
           numGpu:
-            currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu
+            currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu,
+       numPredict: currentChatModelSettings?.numPredict ?? userDefaultModelSettings?.numPredict,
+
         })
         const response = await questionOllama.invoke(promptForQuestion)
         query = response.content.toString()
@@ -964,7 +974,9 @@ export const useMessage = () => {
         currentChatModelSettings?.numCtx ?? userDefaultModelSettings?.numCtx,
       seed: currentChatModelSettings?.seed,
       numGpu:
-        currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu
+        currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu,
+       numPredict: currentChatModelSettings?.numPredict ?? userDefaultModelSettings?.numPredict,
+
     })
 
     let newMessage: Message[] = []

src/hooks/useMessageOption.tsx

@@ -122,7 +122,9 @@ export const useMessageOption = () => {
         currentChatModelSettings?.numCtx ?? userDefaultModelSettings?.numCtx,
       seed: currentChatModelSettings?.seed,
       numGpu:
-        currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu
+        currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu,
+       numPredict: currentChatModelSettings?.numPredict ?? userDefaultModelSettings?.numPredict,
+
     })
 
     let newMessage: Message[] = []
@@ -197,7 +199,9 @@ export const useMessageOption = () => {
             userDefaultModelSettings?.numCtx,
           seed: currentChatModelSettings?.seed,
           numGpu:
-            currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu
+            currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu,
+       numPredict: currentChatModelSettings?.numPredict ?? userDefaultModelSettings?.numPredict,
+
         })
         const response = await questionOllama.invoke(promptForQuestion)
         query = response.content.toString()
@@ -381,7 +385,8 @@ export const useMessageOption = () => {
         currentChatModelSettings?.numCtx ?? userDefaultModelSettings?.numCtx,
       seed: currentChatModelSettings?.seed,
       numGpu:
-        currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu
+        currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu,
+       numPredict: currentChatModelSettings?.numPredict ?? userDefaultModelSettings?.numPredict,
     })
 
     let newMessage: Message[] = []
@@ -616,7 +621,9 @@ export const useMessageOption = () => {
         currentChatModelSettings?.numCtx ?? userDefaultModelSettings?.numCtx,
       seed: currentChatModelSettings?.seed,
       numGpu:
-        currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu
+        currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu,
+       numPredict: currentChatModelSettings?.numPredict ?? userDefaultModelSettings?.numPredict,
+
     })
 
     let newMessage: Message[] = []
@@ -707,7 +714,9 @@ export const useMessageOption = () => {
             userDefaultModelSettings?.numCtx,
           seed: currentChatModelSettings?.seed,
           numGpu:
-            currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu
+            currentChatModelSettings?.numGpu ?? userDefaultModelSettings?.numGpu,
+       numPredict: currentChatModelSettings?.numPredict ?? userDefaultModelSettings?.numPredict,
+
         })
         const response = await questionOllama.invoke(promptForQuestion)
         query = response.content.toString()

src/models/index.ts

@@ -13,7 +13,8 @@ export const pageAssistModel = async ({
   topP,
   numCtx,
   seed,
-  numGpu
+  numGpu,
+  numPredict,
 }: {
   model: string
   baseUrl: string
@@ -24,12 +25,13 @@ export const pageAssistModel = async ({
   numCtx?: number
   seed?: number
   numGpu?: number
+  numPredict?: number
 }) => {
 
   if (model === "chrome::gemini-nano::page-assist") {
     return new ChatChromeAI({
       temperature,
-      topK
+      topK,
     })
   }
 
@@ -46,6 +48,7 @@ export const pageAssistModel = async ({
       openAIApiKey: providerInfo.apiKey || "temp",
       temperature,
       topP,
+      maxTokens: numPredict,
       configuration: {
         apiKey: providerInfo.apiKey || "temp",
         baseURL: providerInfo.baseUrl || "",
@@ -64,7 +67,8 @@ export const pageAssistModel = async ({
     numCtx,
     seed,
     model,
-    numGpu
+    numGpu,
+    numPredict
   })