update mini-orca 3b to gguf2, license

Signed-off-by: Aaron Miller <apage43@ninjawhale.com>

README.md

@@ -92,6 +92,17 @@ Example tags: `backend`, `bindings`, `python-bindings`, `documentation`, etc.
 <a href="https://s3.amazonaws.com/static.nomic.ai/gpt4all/2023_GPT4All_Technical_Report.pdf">:green_book: Technical Report 1: GPT4All</a>
 </p>
 
+## Star History
+
+<a href="https://star-history.com/#nomic-ai/gpt4all&Date">
+  <picture>
+    <source media="(prefers-color-scheme: dark)" srcset="https://api.star-history.com/svg?repos=nomic-ai/gpt4all&type=Date&theme=dark" />
+    <source media="(prefers-color-scheme: light)" srcset="https://api.star-history.com/svg?repos=nomic-ai/gpt4all&type=Date" />
+    <img alt="Star History Chart" src="https://api.star-history.com/svg?repos=nomic-ai/gpt4all&type=Date" />
+  </picture>
+</a>
+
+
 ## Citation
 
 If you utilize this repository, models or data in a downstream project, please consider citing it with:

gpt4all-backend/llama.cpp-mainline

@@ -1 +1 @@
-Subproject commit 3742085b0429cbe0ede49bcb9f891e4a5e25a724
+Subproject commit 500689ad356a81a471a7fb68cc70f7aee5a5f56e

gpt4all-backend/llama.cpp.cmake

@@ -242,6 +242,8 @@ if (LLAMA_KOMPUTE)
           kompute/op_mul_mat_mat_f16.comp
           kompute/op_mul_mat_mat_q8_0.comp
           kompute/op_mul_mat_mat_q4_0.comp
+          kompute/op_mul_mat_mat_q4_1.comp
+          kompute/op_mul_mat_mat_q6_k.comp
           kompute/op_mul_mat_f16.comp
           kompute/op_mul_mat_q8_0.comp
           kompute/op_mul_mat_q4_0.comp
@@ -276,6 +278,8 @@ if (LLAMA_KOMPUTE)
           shaderop_mul_mat_mat_f16.h
           shaderop_mul_mat_mat_q8_0.h
           shaderop_mul_mat_mat_q4_0.h
+          shaderop_mul_mat_mat_q4_1.h
+          shaderop_mul_mat_mat_q6_k.h
           shaderop_mul_mat_f16.h
           shaderop_mul_mat_q8_0.h
           shaderop_mul_mat_q4_0.h

gpt4all-bindings/cli/app.py

@@ -4,13 +4,13 @@ The GPT4All CLI is a self-contained script based on the `gpt4all` and `typer` pa
 REPL to communicate with a language model similar to the chat GUI application, but more basic.
 """
 
+import importlib.metadata
 import io
-import pkg_resources  # should be present as a dependency of gpt4all
 import sys
-import typer
-
 from collections import namedtuple
 from typing_extensions import Annotated
+
+import typer
 from gpt4all import GPT4All
 
 
@@ -79,7 +79,7 @@ def repl(
 
     use_new_loop = False
     try:
-        version = pkg_resources.Environment()['gpt4all'][0].version
+        version = importlib.metadata.version('gpt4all')
         version_major = int(version.split('.')[0])
         if version_major >= 1:
             use_new_loop = True

gpt4all-bindings/java/src/main/java/com/hexadevlabs/gpt4all/LLModel.java

@@ -8,9 +8,8 @@ import java.io.ByteArrayOutputStream;
 import java.nio.charset.StandardCharsets;
 import java.nio.file.Files;
 import java.nio.file.Path;
-import java.util.HashMap;
-import java.util.List;
-import java.util.Map;
+import java.util.*;
+import java.util.stream.Collectors;
 
 public  class LLModel implements AutoCloseable {
 
@@ -306,6 +305,197 @@ public  class LLModel implements AutoCloseable {
         };
     }
 
+    /**
+     * The array of messages for the conversation.
+     */
+    public static class Messages {
+
+        private final List<PromptMessage> messages = new ArrayList<>();
+
+        public Messages(PromptMessage...messages) {
+            this.messages.addAll(Arrays.asList(messages));
+        }
+
+        public Messages(List<PromptMessage> messages) {
+            this.messages.addAll(messages);
+        }
+
+        public Messages addPromptMessage(PromptMessage promptMessage) {
+            this.messages.add(promptMessage);
+            return this;
+        }
+
+        List<PromptMessage> toList() {
+            return Collections.unmodifiableList(this.messages);
+        }
+
+        List<Map<String, String>> toListMap() {
+            return messages.stream()
+                    .map(PromptMessage::toMap).collect(Collectors.toList());
+        }
+
+    }
+
+    /**
+     * A message in the conversation, identical to OpenAI's chat message.
+     */
+    public static class PromptMessage {
+
+        private static final String ROLE = "role";
+        private static final String CONTENT = "content";
+
+        private final Map<String, String> message = new HashMap<>();
+
+        public PromptMessage() {
+        }
+
+        public PromptMessage(Role role, String content) {
+            addRole(role);
+            addContent(content);
+        }
+
+        public PromptMessage addRole(Role role) {
+            return this.addParameter(ROLE, role.type());
+        }
+
+        public PromptMessage addContent(String content) {
+            return this.addParameter(CONTENT, content);
+        }
+
+        public PromptMessage addParameter(String key, String value) {
+            this.message.put(key, value);
+            return this;
+        }
+
+        public String content() {
+            return this.parameter(CONTENT);
+        }
+
+        public Role role() {
+            String role = this.parameter(ROLE);
+            return Role.from(role);
+        }
+
+        public String parameter(String key) {
+            return this.message.get(key);
+        }
+
+        Map<String, String> toMap() {
+            return Collections.unmodifiableMap(this.message);
+        }
+
+    }
+
+    public enum Role {
+
+        SYSTEM("system"), ASSISTANT("assistant"), USER("user");
+
+        private final String type;
+
+        String type() {
+            return this.type;
+        }
+
+        static Role from(String type) {
+
+            if (type == null) {
+                return null;
+            }
+
+            switch (type) {
+                case "system": return SYSTEM;
+                case "assistant": return ASSISTANT;
+                case "user": return USER;
+                default: throw new IllegalArgumentException(
+                        String.format("You passed %s type but only %s are supported",
+                                type, Arrays.toString(Role.values())
+                        )
+                );
+            }
+        }
+
+        Role(String type) {
+            this.type = type;
+        }
+
+        @Override
+        public String toString() {
+            return type();
+        }
+    }
+
+    /**
+     * The result of the completion, similar to OpenAI's format.
+     */
+    public static class CompletionReturn {
+        private String model;
+        private Usage usage;
+        private Choices choices;
+
+        public CompletionReturn(String model, Usage usage, Choices choices) {
+            this.model = model;
+            this.usage = usage;
+            this.choices = choices;
+        }
+
+        public Choices choices() {
+            return choices;
+        }
+
+        public String model() {
+            return model;
+        }
+
+        public Usage usage() {
+            return usage;
+        }
+    }
+
+    /**
+     * The generated completions.
+     */
+    public static class Choices {
+
+        private final List<CompletionChoice> choices = new ArrayList<>();
+
+        public Choices(List<CompletionChoice> choices) {
+            this.choices.addAll(choices);
+        }
+
+        public Choices(CompletionChoice...completionChoices){
+            this.choices.addAll(Arrays.asList(completionChoices));
+        }
+
+        public Choices addCompletionChoice(CompletionChoice completionChoice) {
+            this.choices.add(completionChoice);
+            return this;
+        }
+
+        public CompletionChoice first() {
+            return this.choices.get(0);
+        }
+
+        public int totalChoices() {
+            return this.choices.size();
+        }
+
+        public CompletionChoice get(int index) {
+            return this.choices.get(index);
+        }
+
+        public List<CompletionChoice> choices() {
+            return Collections.unmodifiableList(choices);
+        }
+    }
+
+    /**
+     * A completion choice, similar to OpenAI's format.
+     */
+    public static class CompletionChoice extends PromptMessage {
+        public CompletionChoice(Role role, String content) {
+            super(role, content);
+        }
+    }
 
     public static class ChatCompletionResponse {
         public String model;
@@ -323,6 +513,41 @@ public  class LLModel implements AutoCloseable {
         // Getters and setters
     }
 
+    public CompletionReturn chatCompletionResponse(Messages messages,
+                                                                GenerationConfig generationConfig) {
+        return chatCompletion(messages, generationConfig, false, false);
+    }
+
+    /**
+     * chatCompletion formats the existing chat conversation into a template to be
+     * easier to process for chat UIs. It is not absolutely necessary as generate method
+     * may be directly used to make generations with gpt models.
+     *
+     * @param messages object to create theMessages to send to GPT model
+     * @param generationConfig How to decode/process the generation.
+     * @param streamToStdOut Send tokens as they are calculated Standard output.
+     * @param outputFullPromptToStdOut Should full prompt built out of messages be sent to Standard output.
+     * @return CompletionReturn contains stats and generated Text.
+     */
+    public CompletionReturn chatCompletion(Messages messages,
+                                                  GenerationConfig generationConfig, boolean streamToStdOut,
+                                                  boolean outputFullPromptToStdOut) {
+
+        String fullPrompt = buildPrompt(messages.toListMap());
+
+        if(outputFullPromptToStdOut)
+            System.out.print(fullPrompt);
+
+        String generatedText = generate(fullPrompt, generationConfig, streamToStdOut);
+
+        final CompletionChoice promptMessage = new CompletionChoice(Role.ASSISTANT, generatedText);
+        final Choices choices = new Choices(promptMessage);
+
+        final Usage usage = getUsage(fullPrompt, generatedText);
+        return new CompletionReturn(this.modelName, usage, choices);
+
+    }
+
     public ChatCompletionResponse chatCompletion(List<Map<String, String>> messages,
                                                               GenerationConfig generationConfig) {
         return chatCompletion(messages, generationConfig, false, false);
@@ -352,19 +577,23 @@ public  class LLModel implements AutoCloseable {
         ChatCompletionResponse  response = new ChatCompletionResponse();
         response.model = this.modelName;
 
-        Usage usage = new Usage();
-        usage.promptTokens = fullPrompt.length();
-        usage.completionTokens = generatedText.length();
-        usage.totalTokens = fullPrompt.length() + generatedText.length();
-        response.usage =  usage;
+        response.usage = getUsage(fullPrompt, generatedText);
 
         Map<String, String> message = new HashMap<>();
         message.put("role", "assistant");
         message.put("content", generatedText);
 
         response.choices = List.of(message);
-
         return response;
+
+    }
+
+    private Usage getUsage(String fullPrompt, String generatedText) {
+        Usage usage = new Usage();
+        usage.promptTokens = fullPrompt.length();
+        usage.completionTokens = generatedText.length();
+        usage.totalTokens = fullPrompt.length() + generatedText.length();
+        return usage;
     }
 
     protected static String buildPrompt(List<Map<String, String>> messages) {

gpt4all-bindings/java/src/test/java/com/hexadevlabs/gpt4all/BasicTests.java

@@ -28,6 +28,33 @@ import static org.mockito.Mockito.*;
 @ExtendWith(MockitoExtension.class)
 public class BasicTests {
 
+    @Test
+    public void simplePromptWithObject(){
+
+        LLModel model = Mockito.spy(new LLModel());
+
+        LLModel.GenerationConfig config =
+                LLModel.config()
+                        .withNPredict(20)
+                        .build();
+
+        // The generate method will return "4"
+        doReturn("4").when( model ).generate(anyString(), eq(config), eq(true));
+
+        LLModel.PromptMessage promptMessage1 = new LLModel.PromptMessage(LLModel.Role.SYSTEM, "You are a helpful assistant");
+        LLModel.PromptMessage promptMessage2 = new LLModel.PromptMessage(LLModel.Role.USER, "Add 2+2");
+
+        LLModel.Messages messages = new LLModel.Messages(promptMessage1, promptMessage2);
+
+        LLModel.CompletionReturn response = model.chatCompletion(
+                messages, config, true, true);
+
+        assertTrue( response.choices().first().content().contains("4") );
+
+        // Verifies the prompt and response are certain length.
+        assertEquals( 224 , response.usage().totalTokens );
+    }
+
     @Test
     public void simplePrompt(){
 

gpt4all-bindings/python/gpt4all/gpt4all.py

@@ -39,7 +39,7 @@ class Embed4All:
         Args:
             n_threads: number of CPU threads used by GPT4All. Default is None, then the number of threads are determined automatically.
         """
-        self.gpt4all = GPT4All(model_name or 'ggml-all-MiniLM-L6-v2-f16.gguf', n_threads=n_threads, **kwargs)
+        self.gpt4all = GPT4All(model_name or 'all-MiniLM-L6-v2-f16.gguf', n_threads=n_threads, **kwargs)
 
     def embed(self, text: str) -> List[float]:
         """

gpt4all-bindings/python/gpt4all/pyllmodel.py

@@ -1,23 +1,27 @@
+import atexit
 import ctypes
+import importlib.resources
 import logging
 import os
 import platform
-from queue import Queue
 import re
 import subprocess
 import sys
 import threading
+from contextlib import ExitStack
+from queue import Queue
 from typing import Callable, Iterable, List
 
-import pkg_resources
-
 logger: logging.Logger = logging.getLogger(__name__)
 
 
-# TODO: provide a config file to make this more robust
-LLMODEL_PATH = os.path.join("llmodel_DO_NOT_MODIFY", "build").replace("\\", "\\\\")
-MODEL_LIB_PATH = str(pkg_resources.resource_filename("gpt4all", LLMODEL_PATH)).replace("\\", "\\\\")
+file_manager = ExitStack()
+atexit.register(file_manager.close)  # clean up files on exit
 
+# TODO: provide a config file to make this more robust
+MODEL_LIB_PATH = file_manager.enter_context(importlib.resources.as_file(
+    importlib.resources.files("gpt4all") / "llmodel_DO_NOT_MODIFY" / "build",
+))
 
 def load_llmodel_library():
     system = platform.system()
@@ -36,9 +40,7 @@ def load_llmodel_library():
 
     llmodel_file = "libllmodel" + "." + c_lib_ext
 
-    llmodel_dir = str(pkg_resources.resource_filename("gpt4all", os.path.join(LLMODEL_PATH, llmodel_file))).replace(
-        "\\", "\\\\"
-    )
+    llmodel_dir = str(MODEL_LIB_PATH / llmodel_file).replace("\\", r"\\")
 
     llmodel_lib = ctypes.CDLL(llmodel_dir)
 
@@ -131,7 +133,7 @@ llmodel.llmodel_set_implementation_search_path.restype = None
 llmodel.llmodel_threadCount.argtypes = [ctypes.c_void_p]
 llmodel.llmodel_threadCount.restype = ctypes.c_int32
 
-llmodel.llmodel_set_implementation_search_path(MODEL_LIB_PATH.encode("utf-8"))
+llmodel.llmodel_set_implementation_search_path(str(MODEL_LIB_PATH).replace("\\", r"\\").encode("utf-8"))
 
 llmodel.llmodel_available_gpu_devices.argtypes = [ctypes.c_void_p, ctypes.c_size_t, ctypes.POINTER(ctypes.c_int32)]
 llmodel.llmodel_available_gpu_devices.restype = ctypes.POINTER(LLModelGPUDevice)

gpt4all-chat/README.md

@@ -32,13 +32,8 @@ One click installers for macOS, Linux, and Windows at https://gpt4all.io
 * Multi-chat - a list of current and past chats and the ability to save/delete/export and switch between
 * Text to speech - have the AI response with voice
 * Speech to text - give the prompt with your voice
-* Python bindings
-* Typescript bindings
 * Plugin support for langchain other developer tools
-* Save your prompt/responses to disk
-* Upload prompt/response manually/automatically to nomic.ai to aid future training runs
-* Syntax highlighting support for programming languages, etc.
-* REST API with a built-in webserver in the chat gui itself with a headless operation mode as well
+* chat gui headless operation mode
 * Advanced settings for changing temperature, topk, etc. (DONE)
 * * Improve the accessibility of the installer for screen reader users
 * YOUR IDEA HERE

gpt4all-chat/chat.cpp

@@ -385,7 +385,11 @@ bool Chat::serialize(QDataStream &stream, int version) const
         stream << m_modelInfo.filename();
     if (version > 2)
         stream << m_collections;
-    if (!m_llmodel->serialize(stream, version, true /*serializeKV*/))
+
+    const bool serializeKV = MySettings::globalInstance()->saveChatsContext();
+    if (version > 5)
+        stream << serializeKV;
+    if (!m_llmodel->serialize(stream, version, serializeKV))
         return false;
     if (!m_chatModel->serialize(stream, version))
         return false;
@@ -413,7 +417,6 @@ bool Chat::deserialize(QDataStream &stream, int version)
     if (!m_modelInfo.id().isEmpty())
         emit modelInfoChanged();
 
-    bool deserializeKV = true; // make this a setting
     bool discardKV = m_modelInfo.id().isEmpty();
 
     // Prior to version 2 gptj models had a bug that fixed the kv_cache to F32 instead of F16 so
@@ -425,6 +428,11 @@ bool Chat::deserialize(QDataStream &stream, int version)
         stream >> m_collections;
         emit collectionListChanged(m_collections);
     }
+
+    bool deserializeKV = true;
+    if (version > 5)
+        stream >> deserializeKV;
+
     m_llmodel->setModelInfo(m_modelInfo);
     if (!m_llmodel->deserialize(stream, version, deserializeKV, discardKV))
         return false;

gpt4all-chat/chat.h

@@ -54,6 +54,8 @@ public:
     }
     ChatModel *chatModel() { return m_chatModel; }
 
+    bool isNewChat() const { return m_name == tr("New Chat") && !m_chatModel->count(); }
+
     Q_INVOKABLE void reset();
     Q_INVOKABLE void processSystemPrompt();
     Q_INVOKABLE bool isModelLoaded() const;

gpt4all-chat/chatlistmodel.cpp

@@ -5,7 +5,7 @@
 #include <QDataStream>
 
 #define CHAT_FORMAT_MAGIC 0xF5D553CC
-#define CHAT_FORMAT_VERSION 5
+#define CHAT_FORMAT_VERSION 6
 
 class MyChatListModel: public ChatListModel { };
 Q_GLOBAL_STATIC(MyChatListModel, chatListModelInstance)
@@ -17,11 +17,10 @@ ChatListModel *ChatListModel::globalInstance()
 ChatListModel::ChatListModel()
     : QAbstractListModel(nullptr)
     , m_newChat(nullptr)
-    , m_dummyChat(nullptr)
     , m_serverChat(nullptr)
     , m_currentChat(nullptr)
 {
-    addDummyChat();
+    addChat();
 
     ChatsRestoreThread *thread = new ChatsRestoreThread;
     connect(thread, &ChatsRestoreThread::chatRestored, this, &ChatListModel::restoreChat);
@@ -59,10 +58,7 @@ void ChatListModel::saveChats()
     for (Chat *chat : m_chats) {
         if (chat == m_serverChat)
             continue;
-        const bool isChatGPT = chat->modelInfo().isChatGPT;
-        if (!isChatGPT && !MySettings::globalInstance()->saveChats())
-            continue;
-        if (isChatGPT && !MySettings::globalInstance()->saveChatGPTChats())
+        if (chat->isNewChat())
             continue;
         toSave.append(chat);
     }
@@ -197,47 +193,47 @@ void ChatsRestoreThread::run()
     });
 
     for (FileInfo &f : files) {
-            QFile file(f.file);
-            bool success = file.open(QIODevice::ReadOnly);
-            if (!success) {
-                qWarning() << "ERROR: Couldn't restore chat from file:" << file.fileName();
+        QFile file(f.file);
+        bool success = file.open(QIODevice::ReadOnly);
+        if (!success) {
+            qWarning() << "ERROR: Couldn't restore chat from file:" << file.fileName();
+            continue;
+        }
+        QDataStream in(&file);
+
+        qint32 version = 0;
+        if (!f.oldFile) {
+            // Read and check the header
+            quint32 magic;
+            in >> magic;
+            if (magic != CHAT_FORMAT_MAGIC) {
+                qWarning() << "ERROR: Chat file has bad magic:" << file.fileName();
                 continue;
             }
-            QDataStream in(&file);
 
-            qint32 version = 0;
-            if (!f.oldFile) {
-                // Read and check the header
-                quint32 magic;
-                in >> magic;
-                if (magic != CHAT_FORMAT_MAGIC) {
-                    qWarning() << "ERROR: Chat file has bad magic:" << file.fileName();
-                    continue;
-                }
-
-                // Read the version
-                in >> version;
-                if (version < 1) {
-                    qWarning() << "ERROR: Chat file has non supported version:" << file.fileName();
-                    continue;
-                }
-
-                if (version <= 1)
-                    in.setVersion(QDataStream::Qt_6_2);
+            // Read the version
+            in >> version;
+            if (version < 1) {
+                qWarning() << "ERROR: Chat file has non supported version:" << file.fileName();
+                continue;
             }
 
-            qDebug() << "deserializing chat" << f.file;
+            if (version <= 1)
+                in.setVersion(QDataStream::Qt_6_2);
+        }
 
-            Chat *chat = new Chat;
-            chat->moveToThread(qApp->thread());
-            if (!chat->deserialize(in, version)) {
-                qWarning() << "ERROR: Couldn't deserialize chat from file:" << file.fileName();
-            } else {
-                emit chatRestored(chat);
-            }
-            if (f.oldFile)
-               file.remove(); // No longer storing in this directory
-            file.close();
+        qDebug() << "deserializing chat" << f.file;
+
+        Chat *chat = new Chat;
+        chat->moveToThread(qApp->thread());
+        if (!chat->deserialize(in, version)) {
+            qWarning() << "ERROR: Couldn't deserialize chat from file:" << file.fileName();
+        } else {
+            emit chatRestored(chat);
+        }
+        if (f.oldFile)
+           file.remove(); // No longer storing in this directory
+        file.close();
     }
 
     qint64 elapsedTime = timer.elapsed();
@@ -249,35 +245,13 @@ void ChatListModel::restoreChat(Chat *chat)
     chat->setParent(this);
     connect(chat, &Chat::nameChanged, this, &ChatListModel::nameChanged);
 
-    if (m_dummyChat) {
-        beginResetModel();
-        m_chats = QList<Chat*>({chat});
-        setCurrentChat(chat);
-        delete m_dummyChat;
-        m_dummyChat = nullptr;
-        endResetModel();
-    } else {
-        beginInsertRows(QModelIndex(), m_chats.size(), m_chats.size());
-        m_chats.append(chat);
-        endInsertRows();
-    }
+    beginInsertRows(QModelIndex(), m_chats.size(), m_chats.size());
+    m_chats.append(chat);
+    endInsertRows();
 }
 
 void ChatListModel::chatsRestoredFinished()
 {
-    if (m_dummyChat) {
-        beginResetModel();
-        Chat *dummy = m_dummyChat;
-        m_dummyChat = nullptr;
-        m_chats.clear();
-        addChat();
-        delete dummy;
-        endResetModel();
-    }
-
-    if (m_chats.isEmpty())
-        addChat();
-
     addServerChat();
 }
 

gpt4all-chat/chatlistmodel.h

@@ -84,7 +84,7 @@ public:
     Q_INVOKABLE void addChat()
     {
         // Don't add a new chat if we already have one
-        if (m_newChat || m_dummyChat)
+        if (m_newChat)
             return;
 
         // Create a new chat pointer and connect it to determine when it is populated
@@ -101,18 +101,6 @@ public:
         setCurrentChat(m_newChat);
     }
 
-    Q_INVOKABLE void addDummyChat()
-    {
-        // Create a new dummy chat pointer and don't connect it
-        m_dummyChat = new Chat(this);
-        beginInsertRows(QModelIndex(), 0, 0);
-        m_chats.prepend(m_dummyChat);
-        endInsertRows();
-        emit countChanged();
-        m_currentChat = m_dummyChat;
-        emit currentChatChanged();
-    }
-
     Q_INVOKABLE void addServerChat()
     {
         // Create a new dummy chat pointer and don't connect it
@@ -252,7 +240,6 @@ private Q_SLOTS:
 
 private:
     Chat* m_newChat;
-    Chat* m_dummyChat;
     Chat* m_serverChat;
     Chat* m_currentChat;
     QList<Chat*> m_chats;

gpt4all-chat/metadata/models2.json

@@ -110,17 +110,17 @@
   },
   {
     "order": "i",
-    "md5sum": "aae346fe095e60139ca39b3fda4ac7ae",
+    "md5sum": "0e769317b90ac30d6e09486d61fefa26",
     "name": "Mini Orca (Small)",
-    "filename": "orca-mini-3b.q4_0.gguf",
-    "filesize": "1928648352",
+    "filename": "orca-mini-3b-gguf2-q4_0.gguf",
+    "filesize": "1979946720",
     "requires": "2.5.0",
     "ramrequired": "4",
     "parameters": "3 billion",
     "quant": "q4_0",
     "type": "OpenLLaMa",
-    "description": "<strong>Small version of new model with novel dataset</strong><br><ul><li>Instruction based<li>Explain tuned datasets<li>Orca Research Paper dataset construction approaches<li>Licensed for commercial use</ul>",
-    "url": "https://gpt4all.io/models/gguf/orca-mini-3b.q4_0.gguf",
+    "description": "<strong>Small version of new model with novel dataset</strong><br><ul><li>Instruction based<li>Explain tuned datasets<li>Orca Research Paper dataset construction approaches<li>Cannot be used commercially</ul>",
+    "url": "https://gpt4all.io/models/gguf/orca-mini-3b-gguf2-q4_0.gguf",
     "promptTemplate": "### User:\n%1\n### Response:\n",
     "systemPrompt": "### System:\nYou are an AI assistant that follows instruction extremely well. Help as much as you can.\n\n"
   },

gpt4all-chat/mysettings.cpp

@@ -10,8 +10,7 @@
 #include <QUrl>
 
 static int      default_threadCount         = std::min(4, (int32_t) std::thread::hardware_concurrency());
-static bool     default_saveChats           = false;
-static bool     default_saveChatGPTChats    = true;
+static bool     default_saveChatsContext    = false;
 static bool     default_serverChat          = false;
 static QString  default_userDefaultModel    = "Application default";
 static bool     default_forceMetal          = false;
@@ -103,8 +102,7 @@ void MySettings::restoreApplicationDefaults()
     setFontSize(default_fontSize);
     setDevice(default_device);
     setThreadCount(default_threadCount);
-    setSaveChats(default_saveChats);
-    setSaveChatGPTChats(default_saveChatGPTChats);
+    setSaveChatsContext(default_saveChatsContext);
     setServerChat(default_serverChat);
     setModelPath(defaultLocalModelsPath());
     setUserDefaultModel(default_userDefaultModel);
@@ -397,40 +395,22 @@ void MySettings::setThreadCount(int c)
     emit threadCountChanged();
 }
 
-bool MySettings::saveChats() const
+bool MySettings::saveChatsContext() const
 {
     QSettings setting;
     setting.sync();
-    return setting.value("saveChats", default_saveChats).toBool();
+    return setting.value("saveChatsContext", default_saveChatsContext).toBool();
 }
 
-void MySettings::setSaveChats(bool b)
+void MySettings::setSaveChatsContext(bool b)
 {
-    if (saveChats() == b)
+    if (saveChatsContext() == b)
         return;
 
     QSettings setting;
-    setting.setValue("saveChats", b);
+    setting.setValue("saveChatsContext", b);
     setting.sync();
-    emit saveChatsChanged();
-}
-
-bool MySettings::saveChatGPTChats() const
-{
-    QSettings setting;
-    setting.sync();
-    return setting.value("saveChatGPTChats", default_saveChatGPTChats).toBool();
-}
-
-void MySettings::setSaveChatGPTChats(bool b)
-{
-    if (saveChatGPTChats() == b)
-        return;
-
-    QSettings setting;
-    setting.setValue("saveChatGPTChats", b);
-    setting.sync();
-    emit saveChatGPTChatsChanged();
+    emit saveChatsContextChanged();
 }
 
 bool MySettings::serverChat() const

gpt4all-chat/mysettings.h

@@ -10,8 +10,7 @@ class MySettings : public QObject
 {
     Q_OBJECT
     Q_PROPERTY(int threadCount READ threadCount WRITE setThreadCount NOTIFY threadCountChanged)
-    Q_PROPERTY(bool saveChats READ saveChats WRITE setSaveChats NOTIFY saveChatsChanged)
-    Q_PROPERTY(bool saveChatGPTChats READ saveChatGPTChats WRITE setSaveChatGPTChats NOTIFY saveChatGPTChatsChanged)
+    Q_PROPERTY(bool saveChatsContext READ saveChatsContext WRITE setSaveChatsContext NOTIFY saveChatsContextChanged)
     Q_PROPERTY(bool serverChat READ serverChat WRITE setServerChat NOTIFY serverChatChanged)
     Q_PROPERTY(QString modelPath READ modelPath WRITE setModelPath NOTIFY modelPathChanged)
     Q_PROPERTY(QString userDefaultModel READ userDefaultModel WRITE setUserDefaultModel NOTIFY userDefaultModelChanged)
@@ -64,10 +63,8 @@ public:
     // Application settings
     int threadCount() const;
     void setThreadCount(int c);
-    bool saveChats() const;
-    void setSaveChats(bool b);
-    bool saveChatGPTChats() const;
-    void setSaveChatGPTChats(bool b);
+    bool saveChatsContext() const;
+    void setSaveChatsContext(bool b);
     bool serverChat() const;
     void setServerChat(bool b);
     QString modelPath() const;
@@ -122,8 +119,7 @@ Q_SIGNALS:
     void promptTemplateChanged(const ModelInfo &model);
     void systemPromptChanged(const ModelInfo &model);
     void threadCountChanged();
-    void saveChatsChanged();
-    void saveChatGPTChatsChanged();
+    void saveChatsContextChanged();
     void serverChatChanged();
     void modelPathChanged();
     void userDefaultModelChanged();

gpt4all-chat/network.cpp

@@ -317,16 +317,6 @@ void Network::sendNetworkToggled(bool isActive)
     sendMixpanelEvent("network_toggled", QVector<KeyValue>{kv});
 }
 
-void Network::sendSaveChatsToggled(bool isActive)
-{
-    if (!MySettings::globalInstance()->networkUsageStatsActive())
-        return;
-    KeyValue kv;
-    kv.key = QString("isActive");
-    kv.value = QJsonValue(isActive);
-    sendMixpanelEvent("savechats_toggled", QVector<KeyValue>{kv});
-}
-
 void Network::sendNewChat(int count)
 {
     if (!MySettings::globalInstance()->networkUsageStatsActive())

gpt4all-chat/network.h

@@ -38,7 +38,6 @@ public Q_SLOTS:
     void sendDownloadFinished(const QString &model, bool success);
     Q_INVOKABLE void sendSettingsDialog();
     Q_INVOKABLE void sendNetworkToggled(bool active);
-    Q_INVOKABLE void sendSaveChatsToggled(bool active);
     Q_INVOKABLE void sendNewChat(int count);
     Q_INVOKABLE void sendRemoveChat();
     Q_INVOKABLE void sendRenameChat();

gpt4all-chat/qml/ApplicationSettings.qml

@@ -234,53 +234,35 @@ MySettingsTab {
             Accessible.description: ToolTip.text
         }
         Label {
-            id: saveChatsLabel
-            text: qsTr("Save chats to disk:")
+            id: saveChatsContextLabel
+            text: qsTr("Save chats context to disk:")
             color: theme.textColor
             font.pixelSize: theme.fontSizeLarge
             Layout.row: 7
             Layout.column: 0
         }
         MyCheckBox {
-            id: saveChatsBox
+            id: saveChatsContextBox
             Layout.row: 7
             Layout.column: 1
-            checked: MySettings.saveChats
+            checked: MySettings.saveChatsContext
             onClicked: {
-                Network.sendSaveChatsToggled(saveChatsBox.checked);
-                MySettings.saveChats = !MySettings.saveChats
+                MySettings.saveChatsContext = !MySettings.saveChatsContext
             }
             ToolTip.text: qsTr("WARNING: Saving chats to disk can be ~2GB per chat")
             ToolTip.visible: hovered
         }
-        Label {
-            id: saveChatGPTChatsLabel
-            text: qsTr("Save ChatGPT chats to disk:")
-            color: theme.textColor
-            font.pixelSize: theme.fontSizeLarge
-            Layout.row: 8
-            Layout.column: 0
-        }
-        MyCheckBox {
-            id: saveChatGPTChatsBox
-            Layout.row: 8
-            Layout.column: 1
-            checked: MySettings.saveChatGPTChats
-            onClicked: {
-                MySettings.saveChatGPTChats = !MySettings.saveChatGPTChats
-            }
-        }
         Label {
             id: serverChatLabel
             text: qsTr("Enable API server:")
             color: theme.textColor
             font.pixelSize: theme.fontSizeLarge
-            Layout.row: 9
+            Layout.row: 8
             Layout.column: 0
         }
         MyCheckBox {
             id: serverChatBox
-            Layout.row: 9
+            Layout.row: 8
             Layout.column: 1
             checked: MySettings.serverChat
             onClicked: {
@@ -290,7 +272,7 @@ MySettingsTab {
             ToolTip.visible: hovered
         }
         Rectangle {
-            Layout.row: 10
+            Layout.row: 9
             Layout.column: 0
             Layout.columnSpan: 3
             Layout.fillWidth: true