Archives/gpt4all
2
0
Fork 0
mirror of https://github.com/nomic-ai/gpt4all synced 2024-11-02 09:40:42 +00:00
Code Issues Packages Projects Releases Wiki Activity

Report the actual device we're using.

Browse Source
This commit is contained in:
Adam Treat 2023-09-14 08:25:37 -04:00
parent cf4eb530ce
commit 1fa67a585c
5 changed files with 20 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
gpt4all-chat/chat.cpp
View File

@ -56,6 +56,7 @@ void Chat::connectLLM()
connect(m_llmodel, &ChatLLM::recalcChanged, this, &Chat::handleRecalculating, Qt::QueuedConnection);
connect(m_llmodel, &ChatLLM::generatedNameChanged, this, &Chat::generatedNameChanged, Qt::QueuedConnection);
connect(m_llmodel, &ChatLLM::reportSpeed, this, &Chat::handleTokenSpeedChanged, Qt::QueuedConnection);
connect(m_llmodel, &ChatLLM::reportDevice, this, &Chat::handleDeviceChanged, Qt::QueuedConnection);
connect(m_llmodel, &ChatLLM::databaseResultsChanged, this, &Chat::handleDatabaseResultsChanged, Qt::QueuedConnection);
connect(m_llmodel, &ChatLLM::modelInfoChanged, this, &Chat::handleModelInfoChanged, Qt::QueuedConnection);
@ -345,6 +346,12 @@ void Chat::handleTokenSpeedChanged(const QString &tokenSpeed)
emit tokenSpeedChanged();
}
void Chat::handleDeviceChanged(const QString &device)
{
m_device = device;
emit deviceChanged();
}
void Chat::handleDatabaseResultsChanged(const QList<ResultInfo> &results)
{
m_databaseResults = results;

5
gpt4all-chat/chat.h
View File

@ -25,6 +25,7 @@ class Chat : public QObject
Q_PROPERTY(QList<QString> collectionList READ collectionList NOTIFY collectionListChanged)
Q_PROPERTY(QString modelLoadingError READ modelLoadingError NOTIFY modelLoadingErrorChanged)
Q_PROPERTY(QString tokenSpeed READ tokenSpeed NOTIFY tokenSpeedChanged);
Q_PROPERTY(QString device READ device NOTIFY deviceChanged);
QML_ELEMENT
QML_UNCREATABLE("Only creatable from c++!")
@ -88,6 +89,7 @@ public:
QString modelLoadingError() const { return m_modelLoadingError; }
QString tokenSpeed() const { return m_tokenSpeed; }
QString device() const { return m_device; }
public Q_SLOTS:
void serverNewPromptResponsePair(const QString &prompt);
@ -115,6 +117,7 @@ Q_SIGNALS:
void isServerChanged();
void collectionListChanged(const QList<QString> &collectionList);
void tokenSpeedChanged();
void deviceChanged();
private Q_SLOTS:
void handleResponseChanged(const QString &response);
@ -125,6 +128,7 @@ private Q_SLOTS:
void handleRecalculating();
void handleModelLoadingError(const QString &error);
void handleTokenSpeedChanged(const QString &tokenSpeed);
void handleDeviceChanged(const QString &device);
void handleDatabaseResultsChanged(const QList<ResultInfo> &results);
void handleModelInfoChanged(const ModelInfo &modelInfo);
void handleModelInstalled();
@ -137,6 +141,7 @@ private:
ModelInfo m_modelInfo;
QString m_modelLoadingError;
QString m_tokenSpeed;
QString m_device;
QString m_response;
QList<QString> m_collections;
ChatModel *m_chatModel;

6
gpt4all-chat/chatllm.cpp
View File

@ -271,22 +271,28 @@ bool ChatLLM::loadModel(const ModelInfo &modelInfo)
MySettings::globalInstance()->setDeviceList(deviceList);
// Pick the best match for the device
QString actualDevice = m_llModelInfo.model->implementation().buildVariant() == "metal" ? "Metal" : "CPU";
const QString requestedDevice = MySettings::globalInstance()->device();
if (requestedDevice != "CPU") {
const size_t requiredMemory = m_llModelInfo.model->requiredMem(filePath.toStdString());
std::vector<LLModel::GPUDevice> availableDevices = m_llModelInfo.model->availableGPUDevices(requiredMemory);
if (!availableDevices.empty() && requestedDevice == "Auto" && availableDevices.front().type == 2 /*a discrete gpu*/) {
m_llModelInfo.model->initializeGPUDevice(availableDevices.front());
actualDevice = QString::fromStdString(availableDevices.front().name);
} else {
for (LLModel::GPUDevice &d : availableDevices) {
if (QString::fromStdString(d.name) == requestedDevice) {
m_llModelInfo.model->initializeGPUDevice(d);
actualDevice = QString::fromStdString(d.name);
break;
}
}
}
}
// Report which device we're actually using
emit reportDevice(actualDevice);
bool success = m_llModelInfo.model->loadModel(filePath.toStdString());
MySettings::globalInstance()->setAttemptModelLoad(QString());
if (!success) {

1
gpt4all-chat/chatllm.h
View File

@ -129,6 +129,7 @@ Q_SIGNALS:
void shouldBeLoadedChanged();
void requestRetrieveFromDB(const QList<QString> &collections, const QString &text, int retrievalSize, QList<ResultInfo> *results);
void reportSpeed(const QString &speed);
void reportDevice(const QString &device);
void databaseResultsChanged(const QList<ResultInfo>&);
void modelInfoChanged(const ModelInfo &modelInfo);

2
gpt4all-chat/main.qml
View File

@ -1013,7 +1013,7 @@ Window {
anchors.rightMargin: 30
color: theme.mutedTextColor
visible: currentChat.tokenSpeed !== ""
text: qsTr("Speed: ") + currentChat.tokenSpeed + "<br>" + qsTr("Device: ") + MySettings.device
text: qsTr("Speed: ") + currentChat.tokenSpeed + "<br>" + qsTr("Device: ") + currentChat.device
font.pixelSize: theme.fontSizeLarge
}