🇬 refactor: Update default Google Models and Parameters (#2782)

* Update Google default model and parameters

* Update .env.example Vertex AI Models to reflect latest version and deprecate bison family

* Update Vertex AI model list in .env.example

.env.example

@@ -119,7 +119,7 @@ GOOGLE_KEY=user_provided
 # GOOGLE_MODELS=gemini-1.5-flash-latest,gemini-1.0-pro,gemini-1.0-pro-001,gemini-1.0-pro-latest,gemini-1.0-pro-vision-latest,gemini-1.5-pro-latest,gemini-pro,gemini-pro-vision
 
 # Vertex AI
-# GOOGLE_MODELS=gemini-1.5-flash-preview-0514,gemini-1.5-pro-preview-0409,gemini-1.0-pro-vision-001,gemini-pro,gemini-pro-vision,chat-bison,chat-bison-32k,codechat-bison,codechat-bison-32k,text-bison,text-bison-32k,text-unicorn,code-gecko,code-bison,code-bison-32k
+# GOOGLE_MODELS=gemini-1.5-flash-preview-0514,gemini-1.5-pro-preview-0514,gemini-1.0-pro-vision-001,gemini-1.0-pro-002,gemini-1.0-pro-001,gemini-pro-vision,gemini-1.0-pro
 
 # Google Gemini Safety Settings
 # NOTE (Vertex AI): You do not have access to the BLOCK_NONE setting by default.

client/src/components/Endpoints/Settings/Google.tsx

@@ -1,4 +1,3 @@
-import { useEffect } from 'react';
 import TextareaAutosize from 'react-textarea-autosize';
 import { EModelEndpoint, endpointSettings } from 'librechat-data-provider';
 import type { TModelSelectProps, OnInputNumberChange } from '~/common';
@@ -31,25 +30,6 @@ export default function Settings({ conversation, setOption, models, readonly }:
     maxOutputTokens,
   } = conversation ?? {};
 
-  const isGemini = model?.toLowerCase()?.includes('gemini');
-
-  const maxOutputTokensMax = isGemini
-    ? google.maxOutputTokens.maxGemini
-    : google.maxOutputTokens.max;
-  const maxOutputTokensDefault = isGemini
-    ? google.maxOutputTokens.defaultGemini
-    : google.maxOutputTokens.default;
-
-  useEffect(
-    () => {
-      if (model) {
-        setOption('maxOutputTokens')(Math.min(Number(maxOutputTokens) ?? 0, maxOutputTokensMax));
-      }
-    },
-    // eslint-disable-next-line react-hooks/exhaustive-deps
-    [model],
-  );
-
   const [setMaxContextTokens, maxContextTokensValue] = useDebouncedInput<number | null | undefined>(
     {
       setOption,
@@ -281,15 +261,15 @@ export default function Settings({ conversation, setOption, models, readonly }:
               <Label htmlFor="max-tokens-int" className="text-left text-sm font-medium">
                 {localize('com_endpoint_max_output_tokens')}{' '}
                 <small className="opacity-40">
-                  ({localize('com_endpoint_default_with_num', maxOutputTokensDefault + '')})
+                  ({localize('com_endpoint_default_with_num', google.maxOutputTokens.default + '')})
                 </small>
               </Label>
               <InputNumber
                 id="max-tokens-int"
                 disabled={readonly}
                 value={maxOutputTokens}
-                onChange={(value) => setMaxOutputTokens(value ?? maxOutputTokensDefault)}
+                onChange={(value) => setMaxOutputTokens(Number(value))}
-                max={maxOutputTokensMax}
+                max={google.maxOutputTokens.max}
                 min={google.maxOutputTokens.min}
                 step={google.maxOutputTokens.step}
                 controls={false}
@@ -304,10 +284,10 @@ export default function Settings({ conversation, setOption, models, readonly }:
             </div>
             <Slider
               disabled={readonly}
-              value={[maxOutputTokens ?? maxOutputTokensDefault]}
+              value={[maxOutputTokens ?? google.maxOutputTokens.default]}
               onValueChange={(value) => setMaxOutputTokens(value[0])}
-              doubleClickHandler={() => setMaxOutputTokens(maxOutputTokensDefault)}
+              doubleClickHandler={() => setMaxOutputTokens(google.maxOutputTokens.default)}
-              max={maxOutputTokensMax}
+              max={google.maxOutputTokens.max}
               min={google.maxOutputTokens.min}
               step={google.maxOutputTokens.step}
               className="flex h-4 w-full"

packages/data-provider/src/generate.ts

@@ -555,18 +555,6 @@ export const generateGoogleSchema = (customGoogle: GoogleSettings) => {
       maxContextTokens: true,
     })
     .transform((obj) => {
-      const isGemini = obj?.model?.toLowerCase()?.includes('gemini');
-
-      const maxOutputTokensMax = isGemini
-        ? defaults.maxOutputTokens.maxGemini
-        : defaults.maxOutputTokens.max;
-      const maxOutputTokensDefault = isGemini
-        ? defaults.maxOutputTokens.defaultGemini
-        : defaults.maxOutputTokens.default;
-
-      let maxOutputTokens = obj.maxOutputTokens ?? maxOutputTokensDefault;
-      maxOutputTokens = Math.min(maxOutputTokens, maxOutputTokensMax);
-
       return {
         ...obj,
         model: obj.model ?? defaults.model.default,
@@ -574,7 +562,7 @@ export const generateGoogleSchema = (customGoogle: GoogleSettings) => {
         promptPrefix: obj.promptPrefix ?? null,
         examples: obj.examples ?? [{ input: { content: '' }, output: { content: '' } }],
         temperature: obj.temperature ?? defaults.temperature.default,
-        maxOutputTokens,
+        maxOutputTokens: obj.maxOutputTokens ?? defaults.maxOutputTokens.default,
         topP: obj.topP ?? defaults.topP.default,
         topK: obj.topK ?? defaults.topK.default,
         maxContextTokens: obj.maxContextTokens ?? undefined,

packages/data-provider/src/schemas.ts

@@ -128,27 +128,25 @@ export const openAISettings = {
 
 export const googleSettings = {
   model: {
-    default: 'chat-bison',
+    default: 'gemini-1.5-flash-latest',
   },
   maxOutputTokens: {
     min: 1,
-    max: 2048,
+    max: 8192,
     step: 1,
-    default: 1024,
+    default: 8192,
-    maxGemini: 8192,
-    defaultGemini: 8192,
   },
   temperature: {
     min: 0,
-    max: 1,
+    max: 2,
     step: 0.01,
-    default: 0.2,
+    default: 1,
   },
   topP: {
     min: 0,
     max: 1,
     step: 0.01,
-    default: 0.8,
+    default: 0.95,
   },
   topK: {
     min: 1,
@@ -484,18 +482,6 @@ export const googleSchema = tConversationSchema
     maxContextTokens: true,
   })
   .transform((obj) => {
-    const isGemini = obj?.model?.toLowerCase()?.includes('gemini');
-
-    const maxOutputTokensMax = isGemini
-      ? google.maxOutputTokens.maxGemini
-      : google.maxOutputTokens.max;
-    const maxOutputTokensDefault = isGemini
-      ? google.maxOutputTokens.defaultGemini
-      : google.maxOutputTokens.default;
-
-    let maxOutputTokens = obj.maxOutputTokens ?? maxOutputTokensDefault;
-    maxOutputTokens = Math.min(maxOutputTokens, maxOutputTokensMax);
-
     return {
       ...obj,
       model: obj.model ?? google.model.default,
@@ -503,7 +489,7 @@ export const googleSchema = tConversationSchema
       promptPrefix: obj.promptPrefix ?? null,
       examples: obj.examples ?? [{ input: { content: '' }, output: { content: '' } }],
       temperature: obj.temperature ?? google.temperature.default,
-      maxOutputTokens,
+      maxOutputTokens: obj.maxOutputTokens ?? google.maxOutputTokens.default,
       topP: obj.topP ?? google.topP.default,
       topK: obj.topK ?? google.topK.default,
       iconURL: obj.iconURL ?? undefined,