Merge branch 'master' into 1347-human-readable-confluence-url

collector/utils/WhisperProviders/localWhisper.js

@@ -1,19 +1,23 @@
 const fs = require("fs");
 const path = require("path");
 const { v4 } = require("uuid");
+const defaultWhisper = "Xenova/whisper-small"; // Model Card: https://huggingface.co/Xenova/whisper-small
+const fileSize = {
+  "Xenova/whisper-small": "250mb",
+  "Xenova/whisper-large": "1.56GB",
+};
 
 class LocalWhisper {
-  constructor() {
+  constructor({ options }) {
-    // Model Card: https://huggingface.co/Xenova/whisper-small
+    this.model = options?.WhisperModelPref ?? defaultWhisper;
-    this.model = "Xenova/whisper-small";
+    this.fileSize = fileSize[this.model];
     this.cacheDir = path.resolve(
       process.env.STORAGE_DIR
         ? path.resolve(process.env.STORAGE_DIR, `models`)
         : path.resolve(__dirname, `../../../server/storage/models`)
     );
 
-    this.modelPath = path.resolve(this.cacheDir, "Xenova", "whisper-small");
+    this.modelPath = path.resolve(this.cacheDir, ...this.model.split("/"));
-
     // Make directory when it does not exist in existing installations
     if (!fs.existsSync(this.cacheDir))
       fs.mkdirSync(this.cacheDir, { recursive: true });
@@ -104,7 +108,7 @@ class LocalWhisper {
   async client() {
     if (!fs.existsSync(this.modelPath)) {
       this.#log(
-        `The native whisper model has never been run and will be downloaded right now. Subsequent runs will be faster. (~250MB)`
+        `The native whisper model has never been run and will be downloaded right now. Subsequent runs will be faster. (~${this.fileSize})`
       );
     }
 

docker/.env.example

@@ -124,6 +124,10 @@ GID='1000'
 # COHERE_API_KEY=
 # EMBEDDING_MODEL_PREF='embed-english-v3.0'
 
+# EMBEDDING_ENGINE='voyageai'
+# VOYAGEAI_API_KEY=
+# EMBEDDING_MODEL_PREF='voyage-large-2-instruct'
+
 ###########################################
 ######## Vector Database Selection ########
 ###########################################

frontend/.gitignore

@@ -9,10 +9,8 @@ lerna-debug.log*
 
 node_modules
 dist
-lib
 dist-ssr
 *.local
-!frontend/components/lib
 
 # Editor directories and files
 .vscode/*

frontend/src/components/EmbeddingSelection/VoyageAiOptions/index.jsx

@@ -0,0 +1,50 @@
+export default function VoyageAiOptions({ settings }) {
+  return (
+    <div className="w-full flex flex-col gap-y-4">
+      <div className="w-full flex items-center gap-4">
+        <div className="flex flex-col w-60">
+          <label className="text-white text-sm font-semibold block mb-4">
+            API Key
+          </label>
+          <input
+            type="password"
+            name="VoyageAiApiKey"
+            className="bg-zinc-900 text-white placeholder:text-white/20 text-sm rounded-lg focus:border-white block w-full p-2.5"
+            placeholder="Voyage AI API Key"
+            defaultValue={settings?.VoyageAiApiKey ? "*".repeat(20) : ""}
+            required={true}
+            autoComplete="off"
+            spellCheck={false}
+          />
+        </div>
+        <div className="flex flex-col w-60">
+          <label className="text-white text-sm font-semibold block mb-4">
+            Model Preference
+          </label>
+          <select
+            name="EmbeddingModelPref"
+            required={true}
+            defaultValue={settings?.EmbeddingModelPref}
+            className="bg-zinc-900 border-gray-500 text-white text-sm rounded-lg block w-full p-2.5"
+          >
+            <optgroup label="Available embedding models">
+              {[
+                "voyage-large-2-instruct",
+                "voyage-law-2",
+                "voyage-code-2",
+                "voyage-large-2",
+                "voyage-2",
+              ].map((model) => {
+                return (
+                  <option key={model} value={model}>
+                    {model}
+                  </option>
+                );
+              })}
+            </optgroup>
+          </select>
+        </div>
+      </div>
+    </div>
+  );
+}

frontend/src/components/Footer/index.jsx

@@ -14,6 +14,8 @@ import {
 import React, { useEffect, useState } from "react";
 import SettingsButton from "../SettingsButton";
 import { isMobile } from "react-device-detect";
+import { Tooltip } from "react-tooltip";
+import { v4 } from "uuid";
 
 export const MAX_ICONS = 3;
 export const ICON_COMPONENTS = {
@@ -47,36 +49,48 @@ export default function Footer() {
     return (
       <div className="flex justify-center mb-2">
         <div className="flex space-x-4">
+          <ToolTipWrapper id="open-github">
             <a
               href={paths.github()}
               target="_blank"
               rel="noreferrer"
               className="transition-all duration-300 p-2 rounded-full text-white bg-sidebar-button hover:bg-menu-item-selected-gradient hover:border-slate-100 hover:border-opacity-50 border-transparent border"
               aria-label="Find us on Github"
+              data-tooltip-id="open-github"
+              data-tooltip-content="View source code on Github"
             >
               <GithubLogo weight="fill" className="h-5 w-5 " />
             </a>
+          </ToolTipWrapper>
+          <ToolTipWrapper id="open-documentation">
             <a
               href={paths.docs()}
               target="_blank"
               rel="noreferrer"
-            className="transition-all duration-300 p-2 rounded-full text-white bg-sidebar-button hover:bg-menu-item-selected-gradient hover:border-slate-100 hover:border-opacity-50 border-transparent border"
+              className="w-fit transition-all duration-300 p-2 rounded-full text-white bg-sidebar-button hover:bg-menu-item-selected-gradient hover:border-slate-100 hover:border-opacity-50 border-transparent border"
               aria-label="Docs"
+              data-tooltip-id="open-documentation"
+              data-tooltip-content="Open AnythingLLM help docs"
             >
               <BookOpen weight="fill" className="h-5 w-5 " />
             </a>
+          </ToolTipWrapper>
+          <ToolTipWrapper id="open-discord">
             <a
               href={paths.discord()}
               target="_blank"
               rel="noreferrer"
               className="transition-all duration-300 p-2 rounded-full text-white bg-sidebar-button hover:bg-menu-item-selected-gradient hover:border-slate-100 hover:border-opacity-50 border-transparent border"
               aria-label="Join our Discord server"
+              data-tooltip-id="open-discord"
+              data-tooltip-content="Join the AnythingLLM Discord"
             >
               <DiscordLogo
                 weight="fill"
                 className="h-5 w-5 stroke-slate-200 group-hover:stroke-slate-200"
               />
             </a>
+          </ToolTipWrapper>
           {!isMobile && <SettingsButton />}
         </div>
       </div>
@@ -105,3 +119,17 @@ export default function Footer() {
     </div>
   );
 }
+
+export function ToolTipWrapper({ id = v4(), children }) {
+  return (
+    <div className="flex w-fit">
+      {children}
+      <Tooltip
+        id={id}
+        place="top"
+        delayShow={300}
+        className="tooltip !text-xs z-99"
+      />
+    </div>
+  );
+}

frontend/src/components/LLMSelection/GenericOpenAiOptions/index.jsx

@@ -1,5 +1,6 @@
 export default function GenericOpenAiOptions({ settings }) {
   return (
+    <div className="flex flex-col gap-y-4">
       <div className="flex gap-4 flex-wrap">
         <div className="flex flex-col w-60">
           <label className="text-white text-sm font-semibold block mb-4">
@@ -45,6 +46,8 @@ export default function GenericOpenAiOptions({ settings }) {
             autoComplete="off"
           />
         </div>
+      </div>
+      <div className="flex gap-x-4 flex-wrap">
         <div className="flex flex-col w-60">
           <label className="text-white text-sm font-semibold block mb-4">
             Token context window
@@ -77,5 +80,6 @@ export default function GenericOpenAiOptions({ settings }) {
           />
         </div>
       </div>
+    </div>
   );
 }

frontend/src/components/Modals/MangeWorkspace/DataConnectors/Connectors/Github/index.jsx

@@ -3,7 +3,7 @@ import System from "@/models/system";
 import showToast from "@/utils/toast";
 import pluralize from "pluralize";
 import { TagsInput } from "react-tag-input-component";
-import { Warning } from "@phosphor-icons/react";
+import { Info, Warning } from "@phosphor-icons/react";
 import { Tooltip } from "react-tooltip";
 
 const DEFAULT_BRANCHES = ["main", "master"];
@@ -92,45 +92,7 @@ export default function GithubOptions() {
                     <p className="font-bold text-white">Github Access Token</p>{" "}
                     <p className="text-xs text-white/50 font-light flex items-center">
                       optional
-                      {!accessToken && (
+                      <PATTooltip accessToken={accessToken} />
-                        <Warning
-                          size={14}
-                          className="ml-1 text-orange-500 cursor-pointer"
-                          data-tooltip-id="access-token-tooltip"
-                          data-tooltip-place="right"
-                        />
-                      )}
-                      <Tooltip
-                        delayHide={300}
-                        id="access-token-tooltip"
-                        className="max-w-xs"
-                        clickable={true}
-                      >
-                        <p className="text-sm">
-                          Without a{" "}
-                          <a
-                            href="https://docs.github.com/en/authentication/keeping-your-account-and-data-secure/managing-your-personal-access-tokens"
-                            rel="noreferrer"
-                            target="_blank"
-                            className="underline"
-                            onClick={(e) => e.stopPropagation()}
-                          >
-                            Personal Access Token
-                          </a>
-                          , the GitHub API may limit the number of files that
-                          can be collected due to rate limits. You can{" "}
-                          <a
-                            href="https://github.com/settings/personal-access-tokens/new"
-                            rel="noreferrer"
-                            target="_blank"
-                            className="underline"
-                            onClick={(e) => e.stopPropagation()}
-                          >
-                            create a temporary Access Token
-                          </a>{" "}
-                          to avoid this issue.
-                        </p>
-                      </Tooltip>
                     </p>
                   </label>
                   <p className="text-xs font-normal text-white/50">
@@ -180,6 +142,7 @@ export default function GithubOptions() {
           </div>
 
           <div className="flex flex-col gap-y-2 w-full pr-10">
+            <PATAlert accessToken={accessToken} />
             <button
               type="submit"
               disabled={loading}
@@ -269,3 +232,78 @@ function GitHubBranchSelection({ repo, accessToken }) {
     </div>
   );
 }
+
+function PATAlert({ accessToken }) {
+  if (!!accessToken) return null;
+  return (
+    <div className="flex flex-col md:flex-row md:items-center gap-x-2 text-white mb-4 bg-blue-800/30 w-fit rounded-lg px-4 py-2">
+      <div className="gap-x-2 flex items-center">
+        <Info className="shrink-0" size={25} />
+        <p className="text-sm">
+          Without filling out the <b>Github Access Token</b> this data connector
+          will only be able to collect the <b>top-level</b> files of the repo
+          due to GitHub's public API rate-limits.
+          <br />
+          <br />
+          <a
+            href="https://github.com/settings/personal-access-tokens/new"
+            rel="noreferrer"
+            target="_blank"
+            className="underline"
+            onClick={(e) => e.stopPropagation()}
+          >
+            {" "}
+            Get a free Personal Access Token with a GitHub account here.
+          </a>
+        </p>
+      </div>
+    </div>
+  );
+}
+
+function PATTooltip({ accessToken }) {
+  if (!!accessToken) return null;
+  return (
+    <>
+      {!accessToken && (
+        <Warning
+          size={14}
+          className="ml-1 text-orange-500 cursor-pointer"
+          data-tooltip-id="access-token-tooltip"
+          data-tooltip-place="right"
+        />
+      )}
+      <Tooltip
+        delayHide={300}
+        id="access-token-tooltip"
+        className="max-w-xs"
+        clickable={true}
+      >
+        <p className="text-sm">
+          Without a{" "}
+          <a
+            href="https://docs.github.com/en/authentication/keeping-your-account-and-data-secure/managing-your-personal-access-tokens"
+            rel="noreferrer"
+            target="_blank"
+            className="underline"
+            onClick={(e) => e.stopPropagation()}
+          >
+            Personal Access Token
+          </a>
+          , the GitHub API may limit the number of files that can be collected
+          due to rate limits. You can{" "}
+          <a
+            href="https://github.com/settings/personal-access-tokens/new"
+            rel="noreferrer"
+            target="_blank"
+            className="underline"
+            onClick={(e) => e.stopPropagation()}
+          >
+            create a temporary Access Token
+          </a>{" "}
+          to avoid this issue.
+        </p>
+      </Tooltip>
+    </>
+  );
+}

frontend/src/components/SettingsButton/index.jsx

@@ -3,6 +3,7 @@ import paths from "@/utils/paths";
 import { ArrowUUpLeft, Wrench } from "@phosphor-icons/react";
 import { Link } from "react-router-dom";
 import { useMatch } from "react-router-dom";
+import { ToolTipWrapper } from "../Footer";
 
 export default function SettingsButton() {
   const isInSettings = !!useMatch("/settings/*");
@@ -12,22 +13,32 @@ export default function SettingsButton() {
 
   if (isInSettings)
     return (
+      <ToolTipWrapper id="go-home">
         <Link
           to={paths.home()}
           className="transition-all duration-300 p-2 rounded-full text-white bg-sidebar-button hover:bg-menu-item-selected-gradient hover:border-slate-100 hover:border-opacity-50 border-transparent border"
           aria-label="Home"
+          data-tooltip-id="go-home"
+          data-tooltip-content="Back to workspaces"
         >
           <ArrowUUpLeft className="h-5 w-5" weight="fill" />
         </Link>
+      </ToolTipWrapper>
     );
 
   return (
+    <ToolTipWrapper id="open-settings">
       <Link
-      to={!!user?.role ? paths.settings.system() : paths.settings.appearance()}
+        to={
+          !!user?.role ? paths.settings.system() : paths.settings.appearance()
+        }
         className="transition-all duration-300 p-2 rounded-full text-white bg-sidebar-button hover:bg-menu-item-selected-gradient hover:border-slate-100 hover:border-opacity-50 border-transparent border"
         aria-label="Settings"
+        data-tooltip-id="open-settings"
+        data-tooltip-content="Open settings"
       >
         <Wrench className="h-5 w-5" weight="fill" />
       </Link>
+    </ToolTipWrapper>
   );
 }

frontend/src/components/SettingsSidebar/index.jsx

@@ -329,7 +329,7 @@ const SidebarOptions = ({ user = null }) => (
     <Option
       href={paths.settings.embedSetup()}
       childLinks={[paths.settings.embedChats()]}
-      btnText="Embedded Chat"
+      btnText="Chat Embed Widgets"
       icon={<CodeBlock className="h-5 w-5 flex-shrink-0" />}
       user={user}
       flex={true}
@@ -338,7 +338,7 @@ const SidebarOptions = ({ user = null }) => (
         <>
           <Option
             href={paths.settings.embedChats()}
-            btnText="Embedded Chat History"
+            btnText="Chat Embed History"
             icon={<Barcode className="h-5 w-5 flex-shrink-0" />}
             user={user}
             flex={true}

frontend/src/components/TranscriptionSelection/NativeTranscriptionOptions/index.jsx

@@ -1,38 +1,89 @@
 import { Gauge } from "@phosphor-icons/react";
-export default function NativeTranscriptionOptions() {
+import { useState } from "react";
+
+export default function NativeTranscriptionOptions({ settings }) {
+  const [model, setModel] = useState(settings?.WhisperModelPref);
+
   return (
     <div className="w-full flex flex-col gap-y-4">
-      <div className="flex flex-col md:flex-row md:items-center gap-x-2 text-white mb-4 bg-blue-800/30 w-fit rounded-lg px-4 py-2">
+      <LocalWarning model={model} />
-        <div className="gap-x-2 flex items-center">
-          <Gauge size={25} />
-          <p className="text-sm">
-            Using the local whisper model on machines with limited RAM or CPU
-            can stall AnythingLLM when processing media files.
-            <br />
-            We recommend at least 2GB of RAM and upload files &lt;10Mb.
-            <br />
-            <br />
-            <i>
-              The built-in model will automatically download on the first use.
-            </i>
-          </p>
-        </div>
-      </div>
       <div className="w-full flex items-center gap-4">
         <div className="flex flex-col w-60">
           <label className="text-white text-sm font-semibold block mb-4">
             Model Selection
           </label>
           <select
-            disabled={true}
+            name="WhisperModelPref"
+            defaultValue={model}
+            onChange={(e) => setModel(e.target.value)}
             className="bg-zinc-900 border-gray-500 text-white text-sm rounded-lg block w-full p-2.5"
           >
-            <option disabled={true} selected={true}>
+            {["Xenova/whisper-small", "Xenova/whisper-large"].map(
-              Xenova/whisper-small
+              (value, i) => {
+                return (
+                  <option key={i} value={value}>
+                    {value}
                   </option>
+                );
+              }
+            )}
           </select>
         </div>
       </div>
     </div>
   );
 }
+
+function LocalWarning({ model }) {
+  switch (model) {
+    case "Xenova/whisper-small":
+      return <WhisperSmall />;
+    case "Xenova/whisper-large":
+      return <WhisperLarge />;
+    default:
+      return <WhisperSmall />;
+  }
+}
+
+function WhisperSmall() {
+  return (
+    <div className="flex flex-col md:flex-row md:items-center gap-x-2 text-white mb-4 bg-blue-800/30 w-fit rounded-lg px-4 py-2">
+      <div className="gap-x-2 flex items-center">
+        <Gauge size={25} />
+        <p className="text-sm">
+          Running the <b>whisper-small</b> model on a machine with limited RAM
+          or CPU can stall AnythingLLM when processing media files.
+          <br />
+          We recommend at least 2GB of RAM and upload files &lt;10Mb.
+          <br />
+          <br />
+          <i>
+            This model will automatically download on the first use. (250mb)
+          </i>
+        </p>
+      </div>
+    </div>
+  );
+}
+
+function WhisperLarge() {
+  return (
+    <div className="flex flex-col md:flex-row md:items-center gap-x-2 text-white mb-4 bg-blue-800/30 w-fit rounded-lg px-4 py-2">
+      <div className="gap-x-2 flex items-center">
+        <Gauge size={25} />
+        <p className="text-sm">
+          Using the <b>whisper-large</b> model on machines with limited RAM or
+          CPU can stall AnythingLLM when processing media files. This model is
+          substantially larger than the whisper-small.
+          <br />
+          We recommend at least 8GB of RAM and upload files &lt;10Mb.
+          <br />
+          <br />
+          <i>
+            This model will automatically download on the first use. (1.56GB)
+          </i>
+        </p>
+      </div>
+    </div>
+  );
+}

frontend/src/components/WorkspaceChat/ChatContainer/PromptInput/index.jsx

@@ -12,6 +12,7 @@ import AvailableAgentsButton, {
 } from "./AgentMenu";
 import TextSizeButton from "./TextSizeMenu";
 import SpeechToText from "./SpeechToText";
+import { Tooltip } from "react-tooltip";
 
 export const PROMPT_INPUT_EVENT = "set_prompt_input";
 export default function PromptInput({
@@ -134,14 +135,25 @@ export default function PromptInput({
               {buttonDisabled ? (
                 <StopGenerationButton />
               ) : (
+                <>
                   <button
                     ref={formRef}
                     type="submit"
                     className="inline-flex justify-center rounded-2xl cursor-pointer text-white/60 hover:text-white group ml-4"
+                    data-tooltip-id="send-prompt"
+                    data-tooltip-content="Send prompt message to workspace"
+                    aria-label="Send prompt message to workspace"
                   >
                     <PaperPlaneRight className="w-7 h-7 my-3" weight="fill" />
                     <span className="sr-only">Send message</span>
                   </button>
+                  <Tooltip
+                    id="send-prompt"
+                    place="bottom"
+                    delayShow={300}
+                    className="tooltip !text-xs z-99"
+                  />
+                </>
               )}
             </div>
             <div className="flex justify-between py-3.5">

frontend/src/pages/GeneralSettings/EmbeddingPreference/index.jsx

@@ -10,6 +10,8 @@ import LocalAiLogo from "@/media/llmprovider/localai.png";
 import OllamaLogo from "@/media/llmprovider/ollama.png";
 import LMStudioLogo from "@/media/llmprovider/lmstudio.png";
 import CohereLogo from "@/media/llmprovider/cohere.png";
+import VoyageAiLogo from "@/media/embeddingprovider/voyageai.png";
+
 import PreLoader from "@/components/Preloader";
 import ChangeWarningModal from "@/components/ChangeWarning";
 import OpenAiOptions from "@/components/EmbeddingSelection/OpenAiOptions";
@@ -19,6 +21,7 @@ import NativeEmbeddingOptions from "@/components/EmbeddingSelection/NativeEmbedd
 import OllamaEmbeddingOptions from "@/components/EmbeddingSelection/OllamaOptions";
 import LMStudioEmbeddingOptions from "@/components/EmbeddingSelection/LMStudioOptions";
 import CohereEmbeddingOptions from "@/components/EmbeddingSelection/CohereOptions";
+import VoyageAiOptions from "@/components/EmbeddingSelection/VoyageAiOptions";
 
 import EmbedderItem from "@/components/EmbeddingSelection/EmbedderItem";
 import { CaretUpDown, MagnifyingGlass, X } from "@phosphor-icons/react";
@@ -78,6 +81,13 @@ const EMBEDDERS = [
     options: (settings) => <CohereEmbeddingOptions settings={settings} />,
     description: "Run powerful embedding models from Cohere.",
   },
+  {
+    name: "Voyage AI",
+    value: "voyageai",
+    logo: VoyageAiLogo,
+    options: (settings) => <VoyageAiOptions settings={settings} />,
+    description: "Run powerful embedding models from Voyage AI.",
+  },
 ];
 
 export default function GeneralEmbeddingPreference() {

frontend/src/pages/GeneralSettings/TranscriptionPreference/index.jsx

@@ -12,6 +12,23 @@ import LLMItem from "@/components/LLMSelection/LLMItem";
 import { CaretUpDown, MagnifyingGlass, X } from "@phosphor-icons/react";
 import CTAButton from "@/components/lib/CTAButton";
 
+const PROVIDERS = [
+  {
+    name: "OpenAI",
+    value: "openai",
+    logo: OpenAiLogo,
+    options: (settings) => <OpenAiWhisperOptions settings={settings} />,
+    description: "Leverage the OpenAI Whisper-large model using your API key.",
+  },
+  {
+    name: "AnythingLLM Built-In",
+    value: "local",
+    logo: AnythingLLMIcon,
+    options: (settings) => <NativeTranscriptionOptions settings={settings} />,
+    description: "Run a built-in whisper model on this instance privately.",
+  },
+];
+
 export default function TranscriptionModelPreference() {
   const [saving, setSaving] = useState(false);
   const [hasChanges, setHasChanges] = useState(false);
@@ -68,24 +85,6 @@ export default function TranscriptionModelPreference() {
     fetchKeys();
   }, []);
 
-  const PROVIDERS = [
-    {
-      name: "OpenAI",
-      value: "openai",
-      logo: OpenAiLogo,
-      options: <OpenAiWhisperOptions settings={settings} />,
-      description:
-        "Leverage the OpenAI Whisper-large model using your API key.",
-    },
-    {
-      name: "AnythingLLM Built-In",
-      value: "local",
-      logo: AnythingLLMIcon,
-      options: <NativeTranscriptionOptions settings={settings} />,
-      description: "Run a built-in whisper model on this instance privately.",
-    },
-  ];
-
   useEffect(() => {
     const filtered = PROVIDERS.filter((provider) =>
       provider.name.toLowerCase().includes(searchQuery.toLowerCase())
@@ -228,7 +227,7 @@ export default function TranscriptionModelPreference() {
                 {selectedProvider &&
                   PROVIDERS.find(
                     (provider) => provider.value === selectedProvider
-                  )?.options}
+                  )?.options(settings)}
               </div>
             </div>
           </form>

frontend/src/pages/OnboardingFlow/Steps/DataHandling/index.jsx

@@ -28,6 +28,8 @@ import LanceDbLogo from "@/media/vectordbs/lancedb.png";
 import WeaviateLogo from "@/media/vectordbs/weaviate.png";
 import QDrantLogo from "@/media/vectordbs/qdrant.png";
 import MilvusLogo from "@/media/vectordbs/milvus.png";
+import VoyageAiLogo from "@/media/embeddingprovider/voyageai.png";
+
 import React, { useState, useEffect } from "react";
 import paths from "@/utils/paths";
 import { useNavigate } from "react-router-dom";
@@ -292,6 +294,13 @@ export const EMBEDDING_ENGINE_PRIVACY = {
     ],
     logo: CohereLogo,
   },
+  voyageai: {
+    name: "Voyage AI",
+    description: [
+      "Data sent to Voyage AI's servers is shared according to the terms of service of voyageai.com.",
+    ],
+    logo: VoyageAiLogo,
+  },
 };
 
 export default function DataHandling({ setHeader, setForwardBtn, setBackBtn }) {

frontend/src/pages/WorkspaceSettings/ChatSettings/ChatTemperatureSettings/index.jsx

@@ -20,19 +20,23 @@ export default function ChatTemperatureSettings({
           LLM Temperature
         </label>
         <p className="text-white text-opacity-60 text-xs font-medium py-1.5">
-          This setting controls how &quot;random&quot; or dynamic your chat
+          This setting controls how &quot;creative&quot; your LLM responses will
-          responses will be.
+          be.
           <br />
-          The higher the number (1.0 maximum) the more random and incoherent.
+          The higher the number the more creative. For some models this can lead
+          to incoherent responses when set too high.
           <br />
-          <i>Recommended: {defaults.temp}</i>
+          <br />
+          <i>
+            Most LLMs have various acceptable ranges of valid values. Consult
+            your LLM provider for that information.
+          </i>
         </p>
       </div>
       <input
         name="openAiTemp"
         type="number"
         min={0.0}
-        max={1.0}
         step={0.1}
         onWheel={(e) => e.target.blur()}
         defaultValue={workspace?.openAiTemp ?? defaults.temp}

frontend/src/pages/WorkspaceSettings/GeneralAppearance/index.jsx

@@ -2,7 +2,6 @@ import Workspace from "@/models/workspace";
 import { castToType } from "@/utils/types";
 import showToast from "@/utils/toast";
 import { useEffect, useRef, useState } from "react";
-import VectorCount from "./VectorCount";
 import WorkspaceName from "./WorkspaceName";
 import SuggestedChatMessages from "./SuggestedChatMessages";
 import DeleteWorkspace from "./DeleteWorkspace";
@@ -51,7 +50,6 @@ export default function GeneralInfo({ slug }) {
         onSubmit={handleUpdate}
         className="w-1/2 flex flex-col gap-y-6"
       >
-        <VectorCount reload={true} workspace={workspace} />
         <WorkspaceName
           key={workspace.slug}
           workspace={workspace}

frontend/src/pages/WorkspaceSettings/VectorDatabase/VectorCount/index.jsx

@@ -28,9 +28,6 @@ export default function VectorCount({ reload, workspace }) {
   return (
     <div>
       <h3 className="input-label">Number of vectors</h3>
-      <p className="text-white text-opacity-60 text-xs font-medium py-1">
-        Total number of vectors in your vector database.
-      </p>
       <p className="text-white text-opacity-60 text-sm font-medium">
         {totalVectors}
       </p>

frontend/src/pages/WorkspaceSettings/VectorDatabase/index.jsx

@@ -6,6 +6,7 @@ import VectorDBIdentifier from "./VectorDBIdentifier";
 import MaxContextSnippets from "./MaxContextSnippets";
 import DocumentSimilarityThreshold from "./DocumentSimilarityThreshold";
 import ResetDatabase from "./ResetDatabase";
+import VectorCount from "./VectorCount";
 
 export default function VectorDatabase({ workspace }) {
   const [hasChanges, setHasChanges] = useState(false);
@@ -38,7 +39,10 @@ export default function VectorDatabase({ workspace }) {
       onSubmit={handleUpdate}
       className="w-1/2 flex flex-col gap-y-6"
     >
+      <div className="flex items-start gap-x-5">
         <VectorDBIdentifier workspace={workspace} />
+        <VectorCount reload={true} workspace={workspace} />
+      </div>
       <MaxContextSnippets workspace={workspace} setHasChanges={setHasChanges} />
       <DocumentSimilarityThreshold
         workspace={workspace}

server/.env.example

@@ -121,6 +121,10 @@ JWT_SECRET="my-random-string-for-seeding" # Please generate random string at lea
 # COHERE_API_KEY=
 # EMBEDDING_MODEL_PREF='embed-english-v3.0'
 
+# EMBEDDING_ENGINE='voyageai'
+# VOYAGEAI_API_KEY=
+# EMBEDDING_MODEL_PREF='voyage-large-2-instruct'
+
 ###########################################
 ######## Vector Database Selection ########
 ###########################################

server/.gitignore

@@ -19,3 +19,6 @@ public/
 documents
 vector-cache
 yarn-error.log
+
+# Local SSL Certs for HTTPS
+sslcert

server/endpoints/api/workspace/index.js

@@ -447,6 +447,76 @@ function apiWorkspaceEndpoints(app) {
     }
   );
 
+  app.post(
+    "/v1/workspace/:slug/update-pin",
+    [validApiKey],
+    async (request, response) => {
+      /*
+      #swagger.tags = ['Workspaces']
+      #swagger.description = 'Add or remove pin from a document in a workspace by its unique slug.'
+      #swagger.path = '/workspace/{slug}/update-pin'
+      #swagger.parameters['slug'] = {
+          in: 'path',
+          description: 'Unique slug of workspace to find',
+          required: true,
+          type: 'string'
+      }
+      #swagger.requestBody = {
+        description: 'JSON object with the document path and pin status to update.',
+        required: true,
+        type: 'object',
+        content: {
+          "application/json": {
+            example: {
+              docPath: "custom-documents/my-pdf.pdf-hash.json",
+              pinStatus: true
+            }
+          }
+        }
+      }
+      #swagger.responses[200] = {
+        description: 'OK',
+        content: {
+          "application/json": {
+            schema: {
+              type: 'object',
+              example: {
+                message: 'Pin status updated successfully'
+              }
+            }
+          }
+        }
+      }
+      #swagger.responses[404] = {
+        description: 'Document not found'
+      }
+      #swagger.responses[500] = {
+        description: 'Internal Server Error'
+      }
+      */
+      try {
+        const { slug = null } = request.params;
+        const { docPath, pinStatus = false } = reqBody(request);
+        const workspace = await Workspace.get({ slug });
+
+        const document = await Document.get({
+          workspaceId: workspace.id,
+          docpath: docPath,
+        });
+        if (!document) return response.sendStatus(404).end();
+
+        await Document.update(document.id, { pinned: pinStatus });
+        return response
+          .status(200)
+          .json({ message: "Pin status updated successfully" })
+          .end();
+      } catch (error) {
+        console.error("Error processing the pin status update:", error);
+        return response.status(500).end();
+      }
+    }
+  );
+
   app.post(
     "/v1/workspace/:slug/chat",
     [validApiKey],

server/index.js

@@ -36,7 +36,12 @@ app.use(
   })
 );
 
-require("express-ws")(app);
+if (!!process.env.ENABLE_HTTPS) {
+  bootSSL(app, process.env.SERVER_PORT || 3001);
+} else {
+  require("express-ws")(app); // load WebSockets in non-SSL mode.
+}
+
 app.use("/api", apiRouter);
 systemEndpoints(apiRouter);
 extensionEndpoints(apiRouter);
@@ -109,8 +114,6 @@ app.all("*", function (_, response) {
   response.sendStatus(404);
 });
 
-if (!!process.env.ENABLE_HTTPS) {
+// In non-https mode we need to boot at the end since the server has not yet
-  bootSSL(app, process.env.SERVER_PORT || 3001);
+// started and is `.listen`ing.
-} else {
+if (!process.env.ENABLE_HTTPS) bootHTTP(app, process.env.SERVER_PORT || 3001);
-  bootHTTP(app, process.env.SERVER_PORT || 3001);
-}

server/models/systemSettings.js

@@ -150,6 +150,8 @@ const SystemSettings = {
       // - then it can be shared.
       // --------------------------------------------------------
       WhisperProvider: process.env.WHISPER_PROVIDER || "local",
+      WhisperModelPref:
+        process.env.WHISPER_MODEL_PREF || "Xenova/whisper-small",
 
       // --------------------------------------------------------
       // TTS/STT  Selection Settings & Configs
@@ -424,6 +426,9 @@ const SystemSettings = {
       // Cohere API Keys
       CohereApiKey: !!process.env.COHERE_API_KEY,
       CohereModelPref: process.env.COHERE_MODEL_PREF,
+
+      // VoyageAi API Keys
+      VoyageAiApiKey: !!process.env.VOYAGEAI_API_KEY,
     };
   },
 

server/swagger/openapi.json

@@ -2000,6 +2000,69 @@
         }
       }
     },
+    "/workspace/{slug}/update-pin": {
+      "post": {
+        "tags": [
+          "Workspaces"
+        ],
+        "description": "Add or remove pin from a document in a workspace by its unique slug.",
+        "parameters": [
+          {
+            "name": "slug",
+            "in": "path",
+            "required": true,
+            "schema": {
+              "type": "string"
+            },
+            "description": "Unique slug of workspace to find"
+          },
+          {
+            "name": "Authorization",
+            "in": "header",
+            "schema": {
+              "type": "string"
+            }
+          }
+        ],
+        "responses": {
+          "200": {
+            "description": "OK",
+            "content": {
+              "application/json": {
+                "schema": {
+                  "type": "object",
+                  "example": {
+                    "message": "Pin status updated successfully"
+                  }
+                }
+              }
+            }
+          },
+          "403": {
+            "description": "Forbidden"
+          },
+          "404": {
+            "description": "Document not found"
+          },
+          "500": {
+            "description": "Internal Server Error"
+          }
+        },
+        "requestBody": {
+          "description": "JSON object with the document path and pin status to update.",
+          "required": true,
+          "type": "object",
+          "content": {
+            "application/json": {
+              "example": {
+                "docPath": "custom-documents/my-pdf.pdf-hash.json",
+                "pinStatus": true
+              }
+            }
+          }
+        }
+      }
+    },
     "/v1/workspace/{slug}/chat": {
       "post": {
         "tags": [

server/utils/AiProviders/genericOpenAi/index.js

@@ -2,6 +2,7 @@ const { NativeEmbedder } = require("../../EmbeddingEngines/native");
 const {
   handleDefaultStreamResponseV2,
 } = require("../../helpers/chat/responses");
+const { toValidNumber } = require("../../http");
 
 class GenericOpenAiLLM {
   constructor(embedder = null, modelPreference = null) {
@@ -18,7 +19,9 @@ class GenericOpenAiLLM {
     });
     this.model =
       modelPreference ?? process.env.GENERIC_OPEN_AI_MODEL_PREF ?? null;
-    this.maxTokens = process.env.GENERIC_OPEN_AI_MAX_TOKENS ?? 1024;
+    this.maxTokens = process.env.GENERIC_OPEN_AI_MAX_TOKENS
+      ? toValidNumber(process.env.GENERIC_OPEN_AI_MAX_TOKENS, 1024)
+      : 1024;
     if (!this.model)
       throw new Error("GenericOpenAI must have a valid model set.");
     this.limits = {

server/utils/EmbeddingEngines/voyageAi/index.js

@@ -0,0 +1,45 @@
+class VoyageAiEmbedder {
+  constructor() {
+    if (!process.env.VOYAGEAI_API_KEY)
+      throw new Error("No Voyage AI API key was set.");
+
+    const {
+      VoyageEmbeddings,
+    } = require("@langchain/community/embeddings/voyage");
+    const voyage = new VoyageEmbeddings({
+      apiKey: process.env.VOYAGEAI_API_KEY,
+    });
+
+    this.voyage = voyage;
+    this.model = process.env.EMBEDDING_MODEL_PREF || "voyage-large-2-instruct";
+
+    // Limit of how many strings we can process in a single pass to stay with resource or network limits
+    this.batchSize = 128; // Voyage AI's limit per request is 128 https://docs.voyageai.com/docs/rate-limits#use-larger-batches
+    this.embeddingMaxChunkLength = 4000; // https://docs.voyageai.com/docs/embeddings - assume a token is roughly 4 letters with some padding
+  }
+
+  async embedTextInput(textInput) {
+    const result = await this.voyage.embedDocuments(
+      Array.isArray(textInput) ? textInput : [textInput],
+      { modelName: this.model }
+    );
+    return result || [];
+  }
+
+  async embedChunks(textChunks = []) {
+    try {
+      const embeddings = await this.voyage.embedDocuments(textChunks, {
+        modelName: this.model,
+        batchSize: this.batchSize,
+      });
+      return embeddings;
+    } catch (error) {
+      console.error("Voyage AI Failed to embed:", error);
+      throw error;
+    }
+  }
+}
+
+module.exports = {
+  VoyageAiEmbedder,
+};

server/utils/agents/aibitat/plugins/summarize.js

@@ -1,6 +1,5 @@
 const { Document } = require("../../../../models/documents");
 const { safeJsonParse } = require("../../../http");
-const { validate } = require("uuid");
 const { summarizeContent } = require("../utils/summarize");
 const Provider = require("../providers/ai-provider");
 

server/utils/agents/aibitat/providers/genericOpenAi.js

@@ -2,6 +2,7 @@ const OpenAI = require("openai");
 const Provider = require("./ai-provider.js");
 const InheritMultiple = require("./helpers/classes.js");
 const UnTooled = require("./helpers/untooled.js");
+const { toValidNumber } = require("../../../http/index.js");
 
 /**
  * The agent provider for the Generic OpenAI provider.
@@ -24,7 +25,9 @@ class GenericOpenAiProvider extends InheritMultiple([Provider, UnTooled]) {
     this._client = client;
     this.model = model;
     this.verbose = true;
-    this.maxTokens = process.env.GENERIC_OPEN_AI_MAX_TOKENS ?? 1024;
+    this.maxTokens = process.env.GENERIC_OPEN_AI_MAX_TOKENS
+      ? toValidNumber(process.env.GENERIC_OPEN_AI_MAX_TOKENS, 1024)
+      : 1024;
   }
 
   get client() {

server/utils/boot/index.js

@@ -12,16 +12,18 @@ function bootSSL(app, port = 3001) {
     const privateKey = fs.readFileSync(process.env.HTTPS_KEY_PATH);
     const certificate = fs.readFileSync(process.env.HTTPS_CERT_PATH);
     const credentials = { key: privateKey, cert: certificate };
+    const server = https.createServer(credentials, app);
 
-    https
+    server
-      .createServer(credentials, app)
       .listen(port, async () => {
         await setupTelemetry();
         new CommunicationKey(true);
         console.log(`Primary server in HTTPS mode listening on port ${port}`);
       })
       .on("error", catchSigTerms);
-    return app;
+
+    require("express-ws")(app, server); // Apply same certificate + server for WSS connections
+    return { app, server };
   } catch (e) {
     console.error(
       `\x1b[31m[SSL BOOT FAILED]\x1b[0m ${e.message} - falling back to HTTP boot.`,
@@ -46,7 +48,8 @@ function bootHTTP(app, port = 3001) {
       console.log(`Primary server in HTTP mode listening on port ${port}`);
     })
     .on("error", catchSigTerms);
-  return app;
+
+  return { app, server: null };
 }
 
 function catchSigTerms() {

server/utils/chats/embed.js

@@ -29,6 +29,7 @@ async function streamChatWithForEmbed(
 
   const uuid = uuidv4();
   const LLMConnector = getLLMProvider({
+    provider: embed?.workspace?.chatProvider,
     model: chatModel ?? embed.workspace?.chatModel,
   });
   const VectorDb = getVectorDbClass();

server/utils/collectorApi/index.js

@@ -17,6 +17,7 @@ class CollectorApi {
   #attachOptions() {
     return {
       whisperProvider: process.env.WHISPER_PROVIDER || "local",
+      WhisperModelPref: process.env.WHISPER_MODEL_PREF,
       openAiKey: process.env.OPEN_AI_KEY || null,
     };
   }

server/utils/helpers/index.js

@@ -125,6 +125,9 @@ function getEmbeddingEngineSelection() {
     case "cohere":
       const { CohereEmbedder } = require("../EmbeddingEngines/cohere");
       return new CohereEmbedder();
+    case "voyageai":
+      const { VoyageAiEmbedder } = require("../EmbeddingEngines/voyageAi");
+      return new VoyageAiEmbedder();
     default:
       return new NativeEmbedder();
   }

server/utils/helpers/updateENV.js

@@ -350,12 +350,23 @@ const KEY_MAPPING = {
     checks: [isNotEmpty],
   },
 
+  // VoyageAi Options
+  VoyageAiApiKey: {
+    envKey: "VOYAGEAI_API_KEY",
+    checks: [isNotEmpty],
+  },
+
   // Whisper (transcription) providers
   WhisperProvider: {
     envKey: "WHISPER_PROVIDER",
     checks: [isNotEmpty, supportedTranscriptionProvider],
     postUpdate: [],
   },
+  WhisperModelPref: {
+    envKey: "WHISPER_MODEL_PREF",
+    checks: [validLocalWhisper],
+    postUpdate: [],
+  },
 
   // System Settings
   AuthToken: {
@@ -468,6 +479,16 @@ function supportedTTSProvider(input = "") {
   return validSelection ? null : `${input} is not a valid TTS provider.`;
 }
 
+function validLocalWhisper(input = "") {
+  const validSelection = [
+    "Xenova/whisper-small",
+    "Xenova/whisper-large",
+  ].includes(input);
+  return validSelection
+    ? null
+    : `${input} is not a valid Whisper model selection.`;
+}
+
 function supportedLLM(input = "") {
   const validSelection = [
     "openai",
@@ -530,6 +551,7 @@ function supportedEmbeddingModel(input = "") {
     "ollama",
     "lmstudio",
     "cohere",
+    "voyageai",
   ];
   return supported.includes(input)
     ? null

server/utils/http/index.js

@@ -91,6 +91,11 @@ function isValidUrl(urlString = "") {
   return false;
 }
 
+function toValidNumber(number = null, fallback = null) {
+  if (isNaN(Number(number))) return fallback;
+  return Number(number);
+}
+
 module.exports = {
   reqBody,
   multiUserMode,
@@ -101,4 +106,5 @@ module.exports = {
   parseAuthHeader,
   safeJsonParse,
   isValidUrl,
+  toValidNumber,
 };