diff --git a/api/backend/options.go b/api/backend/options.go
index 3266d602..38f56068 100644
--- a/api/backend/options.go
+++ b/api/backend/options.go
@@ -63,6 +63,8 @@ func gRPCModelOpts(c config.Config) *pb.ModelOptions {
 		F16Memory:      c.F16,
 		MLock:          c.MMlock,
 		RopeFreqBase:   c.RopeFreqBase,
+		RopeScaling:    c.RopeScaling,
+		Type:           c.ModelType,
 		RopeFreqScale:  c.RopeFreqScale,
 		NUMA:           c.NUMA,
 		Embeddings:     c.Embeddings,
diff --git a/api/config/config.go b/api/config/config.go
index fed83d7a..1b27b574 100644
--- a/api/config/config.go
+++ b/api/config/config.go
@@ -128,7 +128,9 @@ type LLMConfig struct {
 	Quantization string  `yaml:"quantization"`
 	MMProj       string  `yaml:"mmproj"`
 
-	RopeScaling    string  `yaml:"rope_scaling"`
+	RopeScaling string `yaml:"rope_scaling"`
+	ModelType   string `yaml:"type"`
+
 	YarnExtFactor  float32 `yaml:"yarn_ext_factor"`
 	YarnAttnFactor float32 `yaml:"yarn_attn_factor"`
 	YarnBetaFast   float32 `yaml:"yarn_beta_fast"`
diff --git a/backend/backend.proto b/backend/backend.proto
index dff5ffe7..e9989aec 100644
--- a/backend/backend.proto
+++ b/backend/backend.proto
@@ -134,6 +134,8 @@ message ModelOptions {
   float YarnAttnFactor = 45;
   float YarnBetaFast = 46;
   float YarnBetaSlow = 47;
+
+  string Type = 49;
 }
 
 message Result {
diff --git a/backend/backend_grpc.pb.go b/backend/backend_grpc.pb.go
new file mode 100644
index 00000000..5c97691d
--- /dev/null
+++ b/backend/backend_grpc.pb.go
@@ -0,0 +1,457 @@
+// Code generated by protoc-gen-go-grpc. DO NOT EDIT.
+// versions:
+// - protoc-gen-go-grpc v1.2.0
+// - protoc             v4.23.4
+// source: backend/backend.proto
+
+package proto
+
+import (
+	context "context"
+	grpc "google.golang.org/grpc"
+	codes "google.golang.org/grpc/codes"
+	status "google.golang.org/grpc/status"
+)
+
+// This is a compile-time assertion to ensure that this generated file
+// is compatible with the grpc package it is being compiled against.
+// Requires gRPC-Go v1.32.0 or later.
+const _ = grpc.SupportPackageIsVersion7
+
+// BackendClient is the client API for Backend service.
+//
+// For semantics around ctx use and closing/ending streaming RPCs, please refer to https://pkg.go.dev/google.golang.org/grpc/?tab=doc#ClientConn.NewStream.
+type BackendClient interface {
+	Health(ctx context.Context, in *HealthMessage, opts ...grpc.CallOption) (*Reply, error)
+	Predict(ctx context.Context, in *PredictOptions, opts ...grpc.CallOption) (*Reply, error)
+	LoadModel(ctx context.Context, in *ModelOptions, opts ...grpc.CallOption) (*Result, error)
+	PredictStream(ctx context.Context, in *PredictOptions, opts ...grpc.CallOption) (Backend_PredictStreamClient, error)
+	Embedding(ctx context.Context, in *PredictOptions, opts ...grpc.CallOption) (*EmbeddingResult, error)
+	GenerateImage(ctx context.Context, in *GenerateImageRequest, opts ...grpc.CallOption) (*Result, error)
+	AudioTranscription(ctx context.Context, in *TranscriptRequest, opts ...grpc.CallOption) (*TranscriptResult, error)
+	TTS(ctx context.Context, in *TTSRequest, opts ...grpc.CallOption) (*Result, error)
+	TokenizeString(ctx context.Context, in *PredictOptions, opts ...grpc.CallOption) (*TokenizationResponse, error)
+	Status(ctx context.Context, in *HealthMessage, opts ...grpc.CallOption) (*StatusResponse, error)
+}
+
+type backendClient struct {
+	cc grpc.ClientConnInterface
+}
+
+func NewBackendClient(cc grpc.ClientConnInterface) BackendClient {
+	return &backendClient{cc}
+}
+
+func (c *backendClient) Health(ctx context.Context, in *HealthMessage, opts ...grpc.CallOption) (*Reply, error) {
+	out := new(Reply)
+	err := c.cc.Invoke(ctx, "/backend.Backend/Health", in, out, opts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+func (c *backendClient) Predict(ctx context.Context, in *PredictOptions, opts ...grpc.CallOption) (*Reply, error) {
+	out := new(Reply)
+	err := c.cc.Invoke(ctx, "/backend.Backend/Predict", in, out, opts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+func (c *backendClient) LoadModel(ctx context.Context, in *ModelOptions, opts ...grpc.CallOption) (*Result, error) {
+	out := new(Result)
+	err := c.cc.Invoke(ctx, "/backend.Backend/LoadModel", in, out, opts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+func (c *backendClient) PredictStream(ctx context.Context, in *PredictOptions, opts ...grpc.CallOption) (Backend_PredictStreamClient, error) {
+	stream, err := c.cc.NewStream(ctx, &Backend_ServiceDesc.Streams[0], "/backend.Backend/PredictStream", opts...)
+	if err != nil {
+		return nil, err
+	}
+	x := &backendPredictStreamClient{stream}
+	if err := x.ClientStream.SendMsg(in); err != nil {
+		return nil, err
+	}
+	if err := x.ClientStream.CloseSend(); err != nil {
+		return nil, err
+	}
+	return x, nil
+}
+
+type Backend_PredictStreamClient interface {
+	Recv() (*Reply, error)
+	grpc.ClientStream
+}
+
+type backendPredictStreamClient struct {
+	grpc.ClientStream
+}
+
+func (x *backendPredictStreamClient) Recv() (*Reply, error) {
+	m := new(Reply)
+	if err := x.ClientStream.RecvMsg(m); err != nil {
+		return nil, err
+	}
+	return m, nil
+}
+
+func (c *backendClient) Embedding(ctx context.Context, in *PredictOptions, opts ...grpc.CallOption) (*EmbeddingResult, error) {
+	out := new(EmbeddingResult)
+	err := c.cc.Invoke(ctx, "/backend.Backend/Embedding", in, out, opts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+func (c *backendClient) GenerateImage(ctx context.Context, in *GenerateImageRequest, opts ...grpc.CallOption) (*Result, error) {
+	out := new(Result)
+	err := c.cc.Invoke(ctx, "/backend.Backend/GenerateImage", in, out, opts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+func (c *backendClient) AudioTranscription(ctx context.Context, in *TranscriptRequest, opts ...grpc.CallOption) (*TranscriptResult, error) {
+	out := new(TranscriptResult)
+	err := c.cc.Invoke(ctx, "/backend.Backend/AudioTranscription", in, out, opts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+func (c *backendClient) TTS(ctx context.Context, in *TTSRequest, opts ...grpc.CallOption) (*Result, error) {
+	out := new(Result)
+	err := c.cc.Invoke(ctx, "/backend.Backend/TTS", in, out, opts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+func (c *backendClient) TokenizeString(ctx context.Context, in *PredictOptions, opts ...grpc.CallOption) (*TokenizationResponse, error) {
+	out := new(TokenizationResponse)
+	err := c.cc.Invoke(ctx, "/backend.Backend/TokenizeString", in, out, opts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+func (c *backendClient) Status(ctx context.Context, in *HealthMessage, opts ...grpc.CallOption) (*StatusResponse, error) {
+	out := new(StatusResponse)
+	err := c.cc.Invoke(ctx, "/backend.Backend/Status", in, out, opts...)
+	if err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+// BackendServer is the server API for Backend service.
+// All implementations must embed UnimplementedBackendServer
+// for forward compatibility
+type BackendServer interface {
+	Health(context.Context, *HealthMessage) (*Reply, error)
+	Predict(context.Context, *PredictOptions) (*Reply, error)
+	LoadModel(context.Context, *ModelOptions) (*Result, error)
+	PredictStream(*PredictOptions, Backend_PredictStreamServer) error
+	Embedding(context.Context, *PredictOptions) (*EmbeddingResult, error)
+	GenerateImage(context.Context, *GenerateImageRequest) (*Result, error)
+	AudioTranscription(context.Context, *TranscriptRequest) (*TranscriptResult, error)
+	TTS(context.Context, *TTSRequest) (*Result, error)
+	TokenizeString(context.Context, *PredictOptions) (*TokenizationResponse, error)
+	Status(context.Context, *HealthMessage) (*StatusResponse, error)
+	mustEmbedUnimplementedBackendServer()
+}
+
+// UnimplementedBackendServer must be embedded to have forward compatible implementations.
+type UnimplementedBackendServer struct {
+}
+
+func (UnimplementedBackendServer) Health(context.Context, *HealthMessage) (*Reply, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method Health not implemented")
+}
+func (UnimplementedBackendServer) Predict(context.Context, *PredictOptions) (*Reply, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method Predict not implemented")
+}
+func (UnimplementedBackendServer) LoadModel(context.Context, *ModelOptions) (*Result, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method LoadModel not implemented")
+}
+func (UnimplementedBackendServer) PredictStream(*PredictOptions, Backend_PredictStreamServer) error {
+	return status.Errorf(codes.Unimplemented, "method PredictStream not implemented")
+}
+func (UnimplementedBackendServer) Embedding(context.Context, *PredictOptions) (*EmbeddingResult, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method Embedding not implemented")
+}
+func (UnimplementedBackendServer) GenerateImage(context.Context, *GenerateImageRequest) (*Result, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method GenerateImage not implemented")
+}
+func (UnimplementedBackendServer) AudioTranscription(context.Context, *TranscriptRequest) (*TranscriptResult, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method AudioTranscription not implemented")
+}
+func (UnimplementedBackendServer) TTS(context.Context, *TTSRequest) (*Result, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method TTS not implemented")
+}
+func (UnimplementedBackendServer) TokenizeString(context.Context, *PredictOptions) (*TokenizationResponse, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method TokenizeString not implemented")
+}
+func (UnimplementedBackendServer) Status(context.Context, *HealthMessage) (*StatusResponse, error) {
+	return nil, status.Errorf(codes.Unimplemented, "method Status not implemented")
+}
+func (UnimplementedBackendServer) mustEmbedUnimplementedBackendServer() {}
+
+// UnsafeBackendServer may be embedded to opt out of forward compatibility for this service.
+// Use of this interface is not recommended, as added methods to BackendServer will
+// result in compilation errors.
+type UnsafeBackendServer interface {
+	mustEmbedUnimplementedBackendServer()
+}
+
+func RegisterBackendServer(s grpc.ServiceRegistrar, srv BackendServer) {
+	s.RegisterService(&Backend_ServiceDesc, srv)
+}
+
+func _Backend_Health_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(HealthMessage)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(BackendServer).Health(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: "/backend.Backend/Health",
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(BackendServer).Health(ctx, req.(*HealthMessage))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+func _Backend_Predict_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(PredictOptions)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(BackendServer).Predict(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: "/backend.Backend/Predict",
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(BackendServer).Predict(ctx, req.(*PredictOptions))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+func _Backend_LoadModel_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(ModelOptions)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(BackendServer).LoadModel(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: "/backend.Backend/LoadModel",
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(BackendServer).LoadModel(ctx, req.(*ModelOptions))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+func _Backend_PredictStream_Handler(srv interface{}, stream grpc.ServerStream) error {
+	m := new(PredictOptions)
+	if err := stream.RecvMsg(m); err != nil {
+		return err
+	}
+	return srv.(BackendServer).PredictStream(m, &backendPredictStreamServer{stream})
+}
+
+type Backend_PredictStreamServer interface {
+	Send(*Reply) error
+	grpc.ServerStream
+}
+
+type backendPredictStreamServer struct {
+	grpc.ServerStream
+}
+
+func (x *backendPredictStreamServer) Send(m *Reply) error {
+	return x.ServerStream.SendMsg(m)
+}
+
+func _Backend_Embedding_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(PredictOptions)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(BackendServer).Embedding(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: "/backend.Backend/Embedding",
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(BackendServer).Embedding(ctx, req.(*PredictOptions))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+func _Backend_GenerateImage_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(GenerateImageRequest)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(BackendServer).GenerateImage(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: "/backend.Backend/GenerateImage",
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(BackendServer).GenerateImage(ctx, req.(*GenerateImageRequest))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+func _Backend_AudioTranscription_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(TranscriptRequest)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(BackendServer).AudioTranscription(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: "/backend.Backend/AudioTranscription",
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(BackendServer).AudioTranscription(ctx, req.(*TranscriptRequest))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+func _Backend_TTS_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(TTSRequest)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(BackendServer).TTS(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: "/backend.Backend/TTS",
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(BackendServer).TTS(ctx, req.(*TTSRequest))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+func _Backend_TokenizeString_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(PredictOptions)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(BackendServer).TokenizeString(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: "/backend.Backend/TokenizeString",
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(BackendServer).TokenizeString(ctx, req.(*PredictOptions))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+func _Backend_Status_Handler(srv interface{}, ctx context.Context, dec func(interface{}) error, interceptor grpc.UnaryServerInterceptor) (interface{}, error) {
+	in := new(HealthMessage)
+	if err := dec(in); err != nil {
+		return nil, err
+	}
+	if interceptor == nil {
+		return srv.(BackendServer).Status(ctx, in)
+	}
+	info := &grpc.UnaryServerInfo{
+		Server:     srv,
+		FullMethod: "/backend.Backend/Status",
+	}
+	handler := func(ctx context.Context, req interface{}) (interface{}, error) {
+		return srv.(BackendServer).Status(ctx, req.(*HealthMessage))
+	}
+	return interceptor(ctx, in, info, handler)
+}
+
+// Backend_ServiceDesc is the grpc.ServiceDesc for Backend service.
+// It's only intended for direct use with grpc.RegisterService,
+// and not to be introspected or modified (even as a copy)
+var Backend_ServiceDesc = grpc.ServiceDesc{
+	ServiceName: "backend.Backend",
+	HandlerType: (*BackendServer)(nil),
+	Methods: []grpc.MethodDesc{
+		{
+			MethodName: "Health",
+			Handler:    _Backend_Health_Handler,
+		},
+		{
+			MethodName: "Predict",
+			Handler:    _Backend_Predict_Handler,
+		},
+		{
+			MethodName: "LoadModel",
+			Handler:    _Backend_LoadModel_Handler,
+		},
+		{
+			MethodName: "Embedding",
+			Handler:    _Backend_Embedding_Handler,
+		},
+		{
+			MethodName: "GenerateImage",
+			Handler:    _Backend_GenerateImage_Handler,
+		},
+		{
+			MethodName: "AudioTranscription",
+			Handler:    _Backend_AudioTranscription_Handler,
+		},
+		{
+			MethodName: "TTS",
+			Handler:    _Backend_TTS_Handler,
+		},
+		{
+			MethodName: "TokenizeString",
+			Handler:    _Backend_TokenizeString_Handler,
+		},
+		{
+			MethodName: "Status",
+			Handler:    _Backend_Status_Handler,
+		},
+	},
+	Streams: []grpc.StreamDesc{
+		{
+			StreamName:    "PredictStream",
+			Handler:       _Backend_PredictStream_Handler,
+			ServerStreams: true,
+		},
+	},
+	Metadata: "backend/backend.proto",
+}
diff --git a/backend/python/autogptq/backend_pb2.py b/backend/python/autogptq/backend_pb2.py
index a4a46e04..be6191b1 100644
--- a/backend/python/autogptq/backend_pb2.py
+++ b/backend/python/autogptq/backend_pb2.py
@@ -1,7 +1,6 @@
 # -*- coding: utf-8 -*-
 # Generated by the protocol buffer compiler.  DO NOT EDIT!
 # source: backend.proto
-# Protobuf Python Version: 4.25.0
 """Generated protocol buffer code."""
 from google.protobuf import descriptor as _descriptor
 from google.protobuf import descriptor_pool as _descriptor_pool
@@ -14,16 +13,17 @@ _sym_db = _symbol_database.Default()
 
 
 
-DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xad\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
+DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xbb\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\x12\x0c\n\x04Type\x18\x31 \x01(\t\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
 
 _globals = globals()
 _builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, _globals)
 _builder.BuildTopDescriptorsAndMessages(DESCRIPTOR, 'backend_pb2', _globals)
 if _descriptor._USE_C_DESCRIPTORS == False:
-  _globals['DESCRIPTOR']._options = None
-  _globals['DESCRIPTOR']._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._options = None
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_options = b'8\001'
+
+  DESCRIPTOR._options = None
+  DESCRIPTOR._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._options = None
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._serialized_options = b'8\001'
   _globals['_HEALTHMESSAGE']._serialized_start=26
   _globals['_HEALTHMESSAGE']._serialized_end=41
   _globals['_PREDICTOPTIONS']._serialized_start=44
@@ -31,31 +31,31 @@ if _descriptor._USE_C_DESCRIPTORS == False:
   _globals['_REPLY']._serialized_start=852
   _globals['_REPLY']._serialized_end=876
   _globals['_MODELOPTIONS']._serialized_start=879
-  _globals['_MODELOPTIONS']._serialized_end=1820
-  _globals['_RESULT']._serialized_start=1822
-  _globals['_RESULT']._serialized_end=1864
-  _globals['_EMBEDDINGRESULT']._serialized_start=1866
-  _globals['_EMBEDDINGRESULT']._serialized_end=1903
-  _globals['_TRANSCRIPTREQUEST']._serialized_start=1905
-  _globals['_TRANSCRIPTREQUEST']._serialized_end=1972
-  _globals['_TRANSCRIPTRESULT']._serialized_start=1974
-  _globals['_TRANSCRIPTRESULT']._serialized_end=2052
-  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2054
-  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2143
-  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2146
-  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2361
-  _globals['_TTSREQUEST']._serialized_start=2363
-  _globals['_TTSREQUEST']._serialized_end=2417
-  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2419
-  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2473
-  _globals['_MEMORYUSAGEDATA']._serialized_start=2476
-  _globals['_MEMORYUSAGEDATA']._serialized_end=2618
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2570
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2618
-  _globals['_STATUSRESPONSE']._serialized_start=2621
-  _globals['_STATUSRESPONSE']._serialized_end=2794
-  _globals['_STATUSRESPONSE_STATE']._serialized_start=2727
-  _globals['_STATUSRESPONSE_STATE']._serialized_end=2794
-  _globals['_BACKEND']._serialized_start=2797
-  _globals['_BACKEND']._serialized_end=3425
+  _globals['_MODELOPTIONS']._serialized_end=1834
+  _globals['_RESULT']._serialized_start=1836
+  _globals['_RESULT']._serialized_end=1878
+  _globals['_EMBEDDINGRESULT']._serialized_start=1880
+  _globals['_EMBEDDINGRESULT']._serialized_end=1917
+  _globals['_TRANSCRIPTREQUEST']._serialized_start=1919
+  _globals['_TRANSCRIPTREQUEST']._serialized_end=1986
+  _globals['_TRANSCRIPTRESULT']._serialized_start=1988
+  _globals['_TRANSCRIPTRESULT']._serialized_end=2066
+  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2068
+  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2157
+  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2160
+  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2375
+  _globals['_TTSREQUEST']._serialized_start=2377
+  _globals['_TTSREQUEST']._serialized_end=2431
+  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2433
+  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2487
+  _globals['_MEMORYUSAGEDATA']._serialized_start=2490
+  _globals['_MEMORYUSAGEDATA']._serialized_end=2632
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2584
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2632
+  _globals['_STATUSRESPONSE']._serialized_start=2635
+  _globals['_STATUSRESPONSE']._serialized_end=2808
+  _globals['_STATUSRESPONSE_STATE']._serialized_start=2741
+  _globals['_STATUSRESPONSE_STATE']._serialized_end=2808
+  _globals['_BACKEND']._serialized_start=2811
+  _globals['_BACKEND']._serialized_end=3439
 # @@protoc_insertion_point(module_scope)
diff --git a/backend/python/bark/backend_pb2.py b/backend/python/bark/backend_pb2.py
index a4a46e04..be6191b1 100644
--- a/backend/python/bark/backend_pb2.py
+++ b/backend/python/bark/backend_pb2.py
@@ -1,7 +1,6 @@
 # -*- coding: utf-8 -*-
 # Generated by the protocol buffer compiler.  DO NOT EDIT!
 # source: backend.proto
-# Protobuf Python Version: 4.25.0
 """Generated protocol buffer code."""
 from google.protobuf import descriptor as _descriptor
 from google.protobuf import descriptor_pool as _descriptor_pool
@@ -14,16 +13,17 @@ _sym_db = _symbol_database.Default()
 
 
 
-DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xad\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
+DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xbb\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\x12\x0c\n\x04Type\x18\x31 \x01(\t\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
 
 _globals = globals()
 _builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, _globals)
 _builder.BuildTopDescriptorsAndMessages(DESCRIPTOR, 'backend_pb2', _globals)
 if _descriptor._USE_C_DESCRIPTORS == False:
-  _globals['DESCRIPTOR']._options = None
-  _globals['DESCRIPTOR']._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._options = None
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_options = b'8\001'
+
+  DESCRIPTOR._options = None
+  DESCRIPTOR._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._options = None
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._serialized_options = b'8\001'
   _globals['_HEALTHMESSAGE']._serialized_start=26
   _globals['_HEALTHMESSAGE']._serialized_end=41
   _globals['_PREDICTOPTIONS']._serialized_start=44
@@ -31,31 +31,31 @@ if _descriptor._USE_C_DESCRIPTORS == False:
   _globals['_REPLY']._serialized_start=852
   _globals['_REPLY']._serialized_end=876
   _globals['_MODELOPTIONS']._serialized_start=879
-  _globals['_MODELOPTIONS']._serialized_end=1820
-  _globals['_RESULT']._serialized_start=1822
-  _globals['_RESULT']._serialized_end=1864
-  _globals['_EMBEDDINGRESULT']._serialized_start=1866
-  _globals['_EMBEDDINGRESULT']._serialized_end=1903
-  _globals['_TRANSCRIPTREQUEST']._serialized_start=1905
-  _globals['_TRANSCRIPTREQUEST']._serialized_end=1972
-  _globals['_TRANSCRIPTRESULT']._serialized_start=1974
-  _globals['_TRANSCRIPTRESULT']._serialized_end=2052
-  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2054
-  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2143
-  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2146
-  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2361
-  _globals['_TTSREQUEST']._serialized_start=2363
-  _globals['_TTSREQUEST']._serialized_end=2417
-  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2419
-  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2473
-  _globals['_MEMORYUSAGEDATA']._serialized_start=2476
-  _globals['_MEMORYUSAGEDATA']._serialized_end=2618
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2570
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2618
-  _globals['_STATUSRESPONSE']._serialized_start=2621
-  _globals['_STATUSRESPONSE']._serialized_end=2794
-  _globals['_STATUSRESPONSE_STATE']._serialized_start=2727
-  _globals['_STATUSRESPONSE_STATE']._serialized_end=2794
-  _globals['_BACKEND']._serialized_start=2797
-  _globals['_BACKEND']._serialized_end=3425
+  _globals['_MODELOPTIONS']._serialized_end=1834
+  _globals['_RESULT']._serialized_start=1836
+  _globals['_RESULT']._serialized_end=1878
+  _globals['_EMBEDDINGRESULT']._serialized_start=1880
+  _globals['_EMBEDDINGRESULT']._serialized_end=1917
+  _globals['_TRANSCRIPTREQUEST']._serialized_start=1919
+  _globals['_TRANSCRIPTREQUEST']._serialized_end=1986
+  _globals['_TRANSCRIPTRESULT']._serialized_start=1988
+  _globals['_TRANSCRIPTRESULT']._serialized_end=2066
+  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2068
+  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2157
+  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2160
+  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2375
+  _globals['_TTSREQUEST']._serialized_start=2377
+  _globals['_TTSREQUEST']._serialized_end=2431
+  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2433
+  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2487
+  _globals['_MEMORYUSAGEDATA']._serialized_start=2490
+  _globals['_MEMORYUSAGEDATA']._serialized_end=2632
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2584
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2632
+  _globals['_STATUSRESPONSE']._serialized_start=2635
+  _globals['_STATUSRESPONSE']._serialized_end=2808
+  _globals['_STATUSRESPONSE_STATE']._serialized_start=2741
+  _globals['_STATUSRESPONSE_STATE']._serialized_end=2808
+  _globals['_BACKEND']._serialized_start=2811
+  _globals['_BACKEND']._serialized_end=3439
 # @@protoc_insertion_point(module_scope)
diff --git a/backend/python/coqui/backend_pb2.py b/backend/python/coqui/backend_pb2.py
index a4a46e04..be6191b1 100644
--- a/backend/python/coqui/backend_pb2.py
+++ b/backend/python/coqui/backend_pb2.py
@@ -1,7 +1,6 @@
 # -*- coding: utf-8 -*-
 # Generated by the protocol buffer compiler.  DO NOT EDIT!
 # source: backend.proto
-# Protobuf Python Version: 4.25.0
 """Generated protocol buffer code."""
 from google.protobuf import descriptor as _descriptor
 from google.protobuf import descriptor_pool as _descriptor_pool
@@ -14,16 +13,17 @@ _sym_db = _symbol_database.Default()
 
 
 
-DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xad\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
+DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xbb\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\x12\x0c\n\x04Type\x18\x31 \x01(\t\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
 
 _globals = globals()
 _builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, _globals)
 _builder.BuildTopDescriptorsAndMessages(DESCRIPTOR, 'backend_pb2', _globals)
 if _descriptor._USE_C_DESCRIPTORS == False:
-  _globals['DESCRIPTOR']._options = None
-  _globals['DESCRIPTOR']._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._options = None
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_options = b'8\001'
+
+  DESCRIPTOR._options = None
+  DESCRIPTOR._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._options = None
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._serialized_options = b'8\001'
   _globals['_HEALTHMESSAGE']._serialized_start=26
   _globals['_HEALTHMESSAGE']._serialized_end=41
   _globals['_PREDICTOPTIONS']._serialized_start=44
@@ -31,31 +31,31 @@ if _descriptor._USE_C_DESCRIPTORS == False:
   _globals['_REPLY']._serialized_start=852
   _globals['_REPLY']._serialized_end=876
   _globals['_MODELOPTIONS']._serialized_start=879
-  _globals['_MODELOPTIONS']._serialized_end=1820
-  _globals['_RESULT']._serialized_start=1822
-  _globals['_RESULT']._serialized_end=1864
-  _globals['_EMBEDDINGRESULT']._serialized_start=1866
-  _globals['_EMBEDDINGRESULT']._serialized_end=1903
-  _globals['_TRANSCRIPTREQUEST']._serialized_start=1905
-  _globals['_TRANSCRIPTREQUEST']._serialized_end=1972
-  _globals['_TRANSCRIPTRESULT']._serialized_start=1974
-  _globals['_TRANSCRIPTRESULT']._serialized_end=2052
-  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2054
-  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2143
-  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2146
-  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2361
-  _globals['_TTSREQUEST']._serialized_start=2363
-  _globals['_TTSREQUEST']._serialized_end=2417
-  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2419
-  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2473
-  _globals['_MEMORYUSAGEDATA']._serialized_start=2476
-  _globals['_MEMORYUSAGEDATA']._serialized_end=2618
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2570
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2618
-  _globals['_STATUSRESPONSE']._serialized_start=2621
-  _globals['_STATUSRESPONSE']._serialized_end=2794
-  _globals['_STATUSRESPONSE_STATE']._serialized_start=2727
-  _globals['_STATUSRESPONSE_STATE']._serialized_end=2794
-  _globals['_BACKEND']._serialized_start=2797
-  _globals['_BACKEND']._serialized_end=3425
+  _globals['_MODELOPTIONS']._serialized_end=1834
+  _globals['_RESULT']._serialized_start=1836
+  _globals['_RESULT']._serialized_end=1878
+  _globals['_EMBEDDINGRESULT']._serialized_start=1880
+  _globals['_EMBEDDINGRESULT']._serialized_end=1917
+  _globals['_TRANSCRIPTREQUEST']._serialized_start=1919
+  _globals['_TRANSCRIPTREQUEST']._serialized_end=1986
+  _globals['_TRANSCRIPTRESULT']._serialized_start=1988
+  _globals['_TRANSCRIPTRESULT']._serialized_end=2066
+  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2068
+  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2157
+  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2160
+  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2375
+  _globals['_TTSREQUEST']._serialized_start=2377
+  _globals['_TTSREQUEST']._serialized_end=2431
+  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2433
+  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2487
+  _globals['_MEMORYUSAGEDATA']._serialized_start=2490
+  _globals['_MEMORYUSAGEDATA']._serialized_end=2632
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2584
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2632
+  _globals['_STATUSRESPONSE']._serialized_start=2635
+  _globals['_STATUSRESPONSE']._serialized_end=2808
+  _globals['_STATUSRESPONSE_STATE']._serialized_start=2741
+  _globals['_STATUSRESPONSE_STATE']._serialized_end=2808
+  _globals['_BACKEND']._serialized_start=2811
+  _globals['_BACKEND']._serialized_end=3439
 # @@protoc_insertion_point(module_scope)
diff --git a/backend/python/diffusers/backend_pb2.py b/backend/python/diffusers/backend_pb2.py
index a4a46e04..be6191b1 100644
--- a/backend/python/diffusers/backend_pb2.py
+++ b/backend/python/diffusers/backend_pb2.py
@@ -1,7 +1,6 @@
 # -*- coding: utf-8 -*-
 # Generated by the protocol buffer compiler.  DO NOT EDIT!
 # source: backend.proto
-# Protobuf Python Version: 4.25.0
 """Generated protocol buffer code."""
 from google.protobuf import descriptor as _descriptor
 from google.protobuf import descriptor_pool as _descriptor_pool
@@ -14,16 +13,17 @@ _sym_db = _symbol_database.Default()
 
 
 
-DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xad\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
+DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xbb\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\x12\x0c\n\x04Type\x18\x31 \x01(\t\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
 
 _globals = globals()
 _builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, _globals)
 _builder.BuildTopDescriptorsAndMessages(DESCRIPTOR, 'backend_pb2', _globals)
 if _descriptor._USE_C_DESCRIPTORS == False:
-  _globals['DESCRIPTOR']._options = None
-  _globals['DESCRIPTOR']._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._options = None
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_options = b'8\001'
+
+  DESCRIPTOR._options = None
+  DESCRIPTOR._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._options = None
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._serialized_options = b'8\001'
   _globals['_HEALTHMESSAGE']._serialized_start=26
   _globals['_HEALTHMESSAGE']._serialized_end=41
   _globals['_PREDICTOPTIONS']._serialized_start=44
@@ -31,31 +31,31 @@ if _descriptor._USE_C_DESCRIPTORS == False:
   _globals['_REPLY']._serialized_start=852
   _globals['_REPLY']._serialized_end=876
   _globals['_MODELOPTIONS']._serialized_start=879
-  _globals['_MODELOPTIONS']._serialized_end=1820
-  _globals['_RESULT']._serialized_start=1822
-  _globals['_RESULT']._serialized_end=1864
-  _globals['_EMBEDDINGRESULT']._serialized_start=1866
-  _globals['_EMBEDDINGRESULT']._serialized_end=1903
-  _globals['_TRANSCRIPTREQUEST']._serialized_start=1905
-  _globals['_TRANSCRIPTREQUEST']._serialized_end=1972
-  _globals['_TRANSCRIPTRESULT']._serialized_start=1974
-  _globals['_TRANSCRIPTRESULT']._serialized_end=2052
-  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2054
-  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2143
-  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2146
-  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2361
-  _globals['_TTSREQUEST']._serialized_start=2363
-  _globals['_TTSREQUEST']._serialized_end=2417
-  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2419
-  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2473
-  _globals['_MEMORYUSAGEDATA']._serialized_start=2476
-  _globals['_MEMORYUSAGEDATA']._serialized_end=2618
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2570
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2618
-  _globals['_STATUSRESPONSE']._serialized_start=2621
-  _globals['_STATUSRESPONSE']._serialized_end=2794
-  _globals['_STATUSRESPONSE_STATE']._serialized_start=2727
-  _globals['_STATUSRESPONSE_STATE']._serialized_end=2794
-  _globals['_BACKEND']._serialized_start=2797
-  _globals['_BACKEND']._serialized_end=3425
+  _globals['_MODELOPTIONS']._serialized_end=1834
+  _globals['_RESULT']._serialized_start=1836
+  _globals['_RESULT']._serialized_end=1878
+  _globals['_EMBEDDINGRESULT']._serialized_start=1880
+  _globals['_EMBEDDINGRESULT']._serialized_end=1917
+  _globals['_TRANSCRIPTREQUEST']._serialized_start=1919
+  _globals['_TRANSCRIPTREQUEST']._serialized_end=1986
+  _globals['_TRANSCRIPTRESULT']._serialized_start=1988
+  _globals['_TRANSCRIPTRESULT']._serialized_end=2066
+  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2068
+  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2157
+  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2160
+  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2375
+  _globals['_TTSREQUEST']._serialized_start=2377
+  _globals['_TTSREQUEST']._serialized_end=2431
+  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2433
+  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2487
+  _globals['_MEMORYUSAGEDATA']._serialized_start=2490
+  _globals['_MEMORYUSAGEDATA']._serialized_end=2632
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2584
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2632
+  _globals['_STATUSRESPONSE']._serialized_start=2635
+  _globals['_STATUSRESPONSE']._serialized_end=2808
+  _globals['_STATUSRESPONSE_STATE']._serialized_start=2741
+  _globals['_STATUSRESPONSE_STATE']._serialized_end=2808
+  _globals['_BACKEND']._serialized_start=2811
+  _globals['_BACKEND']._serialized_end=3439
 # @@protoc_insertion_point(module_scope)
diff --git a/backend/python/exllama/backend_pb2.py b/backend/python/exllama/backend_pb2.py
index a4a46e04..be6191b1 100644
--- a/backend/python/exllama/backend_pb2.py
+++ b/backend/python/exllama/backend_pb2.py
@@ -1,7 +1,6 @@
 # -*- coding: utf-8 -*-
 # Generated by the protocol buffer compiler.  DO NOT EDIT!
 # source: backend.proto
-# Protobuf Python Version: 4.25.0
 """Generated protocol buffer code."""
 from google.protobuf import descriptor as _descriptor
 from google.protobuf import descriptor_pool as _descriptor_pool
@@ -14,16 +13,17 @@ _sym_db = _symbol_database.Default()
 
 
 
-DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xad\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
+DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xbb\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\x12\x0c\n\x04Type\x18\x31 \x01(\t\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
 
 _globals = globals()
 _builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, _globals)
 _builder.BuildTopDescriptorsAndMessages(DESCRIPTOR, 'backend_pb2', _globals)
 if _descriptor._USE_C_DESCRIPTORS == False:
-  _globals['DESCRIPTOR']._options = None
-  _globals['DESCRIPTOR']._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._options = None
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_options = b'8\001'
+
+  DESCRIPTOR._options = None
+  DESCRIPTOR._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._options = None
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._serialized_options = b'8\001'
   _globals['_HEALTHMESSAGE']._serialized_start=26
   _globals['_HEALTHMESSAGE']._serialized_end=41
   _globals['_PREDICTOPTIONS']._serialized_start=44
@@ -31,31 +31,31 @@ if _descriptor._USE_C_DESCRIPTORS == False:
   _globals['_REPLY']._serialized_start=852
   _globals['_REPLY']._serialized_end=876
   _globals['_MODELOPTIONS']._serialized_start=879
-  _globals['_MODELOPTIONS']._serialized_end=1820
-  _globals['_RESULT']._serialized_start=1822
-  _globals['_RESULT']._serialized_end=1864
-  _globals['_EMBEDDINGRESULT']._serialized_start=1866
-  _globals['_EMBEDDINGRESULT']._serialized_end=1903
-  _globals['_TRANSCRIPTREQUEST']._serialized_start=1905
-  _globals['_TRANSCRIPTREQUEST']._serialized_end=1972
-  _globals['_TRANSCRIPTRESULT']._serialized_start=1974
-  _globals['_TRANSCRIPTRESULT']._serialized_end=2052
-  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2054
-  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2143
-  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2146
-  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2361
-  _globals['_TTSREQUEST']._serialized_start=2363
-  _globals['_TTSREQUEST']._serialized_end=2417
-  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2419
-  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2473
-  _globals['_MEMORYUSAGEDATA']._serialized_start=2476
-  _globals['_MEMORYUSAGEDATA']._serialized_end=2618
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2570
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2618
-  _globals['_STATUSRESPONSE']._serialized_start=2621
-  _globals['_STATUSRESPONSE']._serialized_end=2794
-  _globals['_STATUSRESPONSE_STATE']._serialized_start=2727
-  _globals['_STATUSRESPONSE_STATE']._serialized_end=2794
-  _globals['_BACKEND']._serialized_start=2797
-  _globals['_BACKEND']._serialized_end=3425
+  _globals['_MODELOPTIONS']._serialized_end=1834
+  _globals['_RESULT']._serialized_start=1836
+  _globals['_RESULT']._serialized_end=1878
+  _globals['_EMBEDDINGRESULT']._serialized_start=1880
+  _globals['_EMBEDDINGRESULT']._serialized_end=1917
+  _globals['_TRANSCRIPTREQUEST']._serialized_start=1919
+  _globals['_TRANSCRIPTREQUEST']._serialized_end=1986
+  _globals['_TRANSCRIPTRESULT']._serialized_start=1988
+  _globals['_TRANSCRIPTRESULT']._serialized_end=2066
+  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2068
+  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2157
+  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2160
+  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2375
+  _globals['_TTSREQUEST']._serialized_start=2377
+  _globals['_TTSREQUEST']._serialized_end=2431
+  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2433
+  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2487
+  _globals['_MEMORYUSAGEDATA']._serialized_start=2490
+  _globals['_MEMORYUSAGEDATA']._serialized_end=2632
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2584
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2632
+  _globals['_STATUSRESPONSE']._serialized_start=2635
+  _globals['_STATUSRESPONSE']._serialized_end=2808
+  _globals['_STATUSRESPONSE_STATE']._serialized_start=2741
+  _globals['_STATUSRESPONSE_STATE']._serialized_end=2808
+  _globals['_BACKEND']._serialized_start=2811
+  _globals['_BACKEND']._serialized_end=3439
 # @@protoc_insertion_point(module_scope)
diff --git a/backend/python/exllama2/backend_pb2.py b/backend/python/exllama2/backend_pb2.py
index a4a46e04..be6191b1 100644
--- a/backend/python/exllama2/backend_pb2.py
+++ b/backend/python/exllama2/backend_pb2.py
@@ -1,7 +1,6 @@
 # -*- coding: utf-8 -*-
 # Generated by the protocol buffer compiler.  DO NOT EDIT!
 # source: backend.proto
-# Protobuf Python Version: 4.25.0
 """Generated protocol buffer code."""
 from google.protobuf import descriptor as _descriptor
 from google.protobuf import descriptor_pool as _descriptor_pool
@@ -14,16 +13,17 @@ _sym_db = _symbol_database.Default()
 
 
 
-DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xad\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
+DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xbb\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\x12\x0c\n\x04Type\x18\x31 \x01(\t\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
 
 _globals = globals()
 _builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, _globals)
 _builder.BuildTopDescriptorsAndMessages(DESCRIPTOR, 'backend_pb2', _globals)
 if _descriptor._USE_C_DESCRIPTORS == False:
-  _globals['DESCRIPTOR']._options = None
-  _globals['DESCRIPTOR']._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._options = None
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_options = b'8\001'
+
+  DESCRIPTOR._options = None
+  DESCRIPTOR._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._options = None
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._serialized_options = b'8\001'
   _globals['_HEALTHMESSAGE']._serialized_start=26
   _globals['_HEALTHMESSAGE']._serialized_end=41
   _globals['_PREDICTOPTIONS']._serialized_start=44
@@ -31,31 +31,31 @@ if _descriptor._USE_C_DESCRIPTORS == False:
   _globals['_REPLY']._serialized_start=852
   _globals['_REPLY']._serialized_end=876
   _globals['_MODELOPTIONS']._serialized_start=879
-  _globals['_MODELOPTIONS']._serialized_end=1820
-  _globals['_RESULT']._serialized_start=1822
-  _globals['_RESULT']._serialized_end=1864
-  _globals['_EMBEDDINGRESULT']._serialized_start=1866
-  _globals['_EMBEDDINGRESULT']._serialized_end=1903
-  _globals['_TRANSCRIPTREQUEST']._serialized_start=1905
-  _globals['_TRANSCRIPTREQUEST']._serialized_end=1972
-  _globals['_TRANSCRIPTRESULT']._serialized_start=1974
-  _globals['_TRANSCRIPTRESULT']._serialized_end=2052
-  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2054
-  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2143
-  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2146
-  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2361
-  _globals['_TTSREQUEST']._serialized_start=2363
-  _globals['_TTSREQUEST']._serialized_end=2417
-  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2419
-  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2473
-  _globals['_MEMORYUSAGEDATA']._serialized_start=2476
-  _globals['_MEMORYUSAGEDATA']._serialized_end=2618
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2570
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2618
-  _globals['_STATUSRESPONSE']._serialized_start=2621
-  _globals['_STATUSRESPONSE']._serialized_end=2794
-  _globals['_STATUSRESPONSE_STATE']._serialized_start=2727
-  _globals['_STATUSRESPONSE_STATE']._serialized_end=2794
-  _globals['_BACKEND']._serialized_start=2797
-  _globals['_BACKEND']._serialized_end=3425
+  _globals['_MODELOPTIONS']._serialized_end=1834
+  _globals['_RESULT']._serialized_start=1836
+  _globals['_RESULT']._serialized_end=1878
+  _globals['_EMBEDDINGRESULT']._serialized_start=1880
+  _globals['_EMBEDDINGRESULT']._serialized_end=1917
+  _globals['_TRANSCRIPTREQUEST']._serialized_start=1919
+  _globals['_TRANSCRIPTREQUEST']._serialized_end=1986
+  _globals['_TRANSCRIPTRESULT']._serialized_start=1988
+  _globals['_TRANSCRIPTRESULT']._serialized_end=2066
+  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2068
+  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2157
+  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2160
+  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2375
+  _globals['_TTSREQUEST']._serialized_start=2377
+  _globals['_TTSREQUEST']._serialized_end=2431
+  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2433
+  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2487
+  _globals['_MEMORYUSAGEDATA']._serialized_start=2490
+  _globals['_MEMORYUSAGEDATA']._serialized_end=2632
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2584
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2632
+  _globals['_STATUSRESPONSE']._serialized_start=2635
+  _globals['_STATUSRESPONSE']._serialized_end=2808
+  _globals['_STATUSRESPONSE_STATE']._serialized_start=2741
+  _globals['_STATUSRESPONSE_STATE']._serialized_end=2808
+  _globals['_BACKEND']._serialized_start=2811
+  _globals['_BACKEND']._serialized_end=3439
 # @@protoc_insertion_point(module_scope)
diff --git a/backend/python/mamba/backend_pb2.py b/backend/python/mamba/backend_pb2.py
index a4a46e04..be6191b1 100644
--- a/backend/python/mamba/backend_pb2.py
+++ b/backend/python/mamba/backend_pb2.py
@@ -1,7 +1,6 @@
 # -*- coding: utf-8 -*-
 # Generated by the protocol buffer compiler.  DO NOT EDIT!
 # source: backend.proto
-# Protobuf Python Version: 4.25.0
 """Generated protocol buffer code."""
 from google.protobuf import descriptor as _descriptor
 from google.protobuf import descriptor_pool as _descriptor_pool
@@ -14,16 +13,17 @@ _sym_db = _symbol_database.Default()
 
 
 
-DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xad\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
+DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xbb\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\x12\x0c\n\x04Type\x18\x31 \x01(\t\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
 
 _globals = globals()
 _builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, _globals)
 _builder.BuildTopDescriptorsAndMessages(DESCRIPTOR, 'backend_pb2', _globals)
 if _descriptor._USE_C_DESCRIPTORS == False:
-  _globals['DESCRIPTOR']._options = None
-  _globals['DESCRIPTOR']._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._options = None
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_options = b'8\001'
+
+  DESCRIPTOR._options = None
+  DESCRIPTOR._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._options = None
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._serialized_options = b'8\001'
   _globals['_HEALTHMESSAGE']._serialized_start=26
   _globals['_HEALTHMESSAGE']._serialized_end=41
   _globals['_PREDICTOPTIONS']._serialized_start=44
@@ -31,31 +31,31 @@ if _descriptor._USE_C_DESCRIPTORS == False:
   _globals['_REPLY']._serialized_start=852
   _globals['_REPLY']._serialized_end=876
   _globals['_MODELOPTIONS']._serialized_start=879
-  _globals['_MODELOPTIONS']._serialized_end=1820
-  _globals['_RESULT']._serialized_start=1822
-  _globals['_RESULT']._serialized_end=1864
-  _globals['_EMBEDDINGRESULT']._serialized_start=1866
-  _globals['_EMBEDDINGRESULT']._serialized_end=1903
-  _globals['_TRANSCRIPTREQUEST']._serialized_start=1905
-  _globals['_TRANSCRIPTREQUEST']._serialized_end=1972
-  _globals['_TRANSCRIPTRESULT']._serialized_start=1974
-  _globals['_TRANSCRIPTRESULT']._serialized_end=2052
-  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2054
-  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2143
-  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2146
-  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2361
-  _globals['_TTSREQUEST']._serialized_start=2363
-  _globals['_TTSREQUEST']._serialized_end=2417
-  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2419
-  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2473
-  _globals['_MEMORYUSAGEDATA']._serialized_start=2476
-  _globals['_MEMORYUSAGEDATA']._serialized_end=2618
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2570
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2618
-  _globals['_STATUSRESPONSE']._serialized_start=2621
-  _globals['_STATUSRESPONSE']._serialized_end=2794
-  _globals['_STATUSRESPONSE_STATE']._serialized_start=2727
-  _globals['_STATUSRESPONSE_STATE']._serialized_end=2794
-  _globals['_BACKEND']._serialized_start=2797
-  _globals['_BACKEND']._serialized_end=3425
+  _globals['_MODELOPTIONS']._serialized_end=1834
+  _globals['_RESULT']._serialized_start=1836
+  _globals['_RESULT']._serialized_end=1878
+  _globals['_EMBEDDINGRESULT']._serialized_start=1880
+  _globals['_EMBEDDINGRESULT']._serialized_end=1917
+  _globals['_TRANSCRIPTREQUEST']._serialized_start=1919
+  _globals['_TRANSCRIPTREQUEST']._serialized_end=1986
+  _globals['_TRANSCRIPTRESULT']._serialized_start=1988
+  _globals['_TRANSCRIPTRESULT']._serialized_end=2066
+  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2068
+  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2157
+  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2160
+  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2375
+  _globals['_TTSREQUEST']._serialized_start=2377
+  _globals['_TTSREQUEST']._serialized_end=2431
+  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2433
+  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2487
+  _globals['_MEMORYUSAGEDATA']._serialized_start=2490
+  _globals['_MEMORYUSAGEDATA']._serialized_end=2632
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2584
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2632
+  _globals['_STATUSRESPONSE']._serialized_start=2635
+  _globals['_STATUSRESPONSE']._serialized_end=2808
+  _globals['_STATUSRESPONSE_STATE']._serialized_start=2741
+  _globals['_STATUSRESPONSE_STATE']._serialized_end=2808
+  _globals['_BACKEND']._serialized_start=2811
+  _globals['_BACKEND']._serialized_end=3439
 # @@protoc_insertion_point(module_scope)
diff --git a/backend/python/petals/backend_pb2.py b/backend/python/petals/backend_pb2.py
index a4a46e04..be6191b1 100644
--- a/backend/python/petals/backend_pb2.py
+++ b/backend/python/petals/backend_pb2.py
@@ -1,7 +1,6 @@
 # -*- coding: utf-8 -*-
 # Generated by the protocol buffer compiler.  DO NOT EDIT!
 # source: backend.proto
-# Protobuf Python Version: 4.25.0
 """Generated protocol buffer code."""
 from google.protobuf import descriptor as _descriptor
 from google.protobuf import descriptor_pool as _descriptor_pool
@@ -14,16 +13,17 @@ _sym_db = _symbol_database.Default()
 
 
 
-DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xad\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
+DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xbb\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\x12\x0c\n\x04Type\x18\x31 \x01(\t\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
 
 _globals = globals()
 _builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, _globals)
 _builder.BuildTopDescriptorsAndMessages(DESCRIPTOR, 'backend_pb2', _globals)
 if _descriptor._USE_C_DESCRIPTORS == False:
-  _globals['DESCRIPTOR']._options = None
-  _globals['DESCRIPTOR']._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._options = None
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_options = b'8\001'
+
+  DESCRIPTOR._options = None
+  DESCRIPTOR._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._options = None
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._serialized_options = b'8\001'
   _globals['_HEALTHMESSAGE']._serialized_start=26
   _globals['_HEALTHMESSAGE']._serialized_end=41
   _globals['_PREDICTOPTIONS']._serialized_start=44
@@ -31,31 +31,31 @@ if _descriptor._USE_C_DESCRIPTORS == False:
   _globals['_REPLY']._serialized_start=852
   _globals['_REPLY']._serialized_end=876
   _globals['_MODELOPTIONS']._serialized_start=879
-  _globals['_MODELOPTIONS']._serialized_end=1820
-  _globals['_RESULT']._serialized_start=1822
-  _globals['_RESULT']._serialized_end=1864
-  _globals['_EMBEDDINGRESULT']._serialized_start=1866
-  _globals['_EMBEDDINGRESULT']._serialized_end=1903
-  _globals['_TRANSCRIPTREQUEST']._serialized_start=1905
-  _globals['_TRANSCRIPTREQUEST']._serialized_end=1972
-  _globals['_TRANSCRIPTRESULT']._serialized_start=1974
-  _globals['_TRANSCRIPTRESULT']._serialized_end=2052
-  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2054
-  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2143
-  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2146
-  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2361
-  _globals['_TTSREQUEST']._serialized_start=2363
-  _globals['_TTSREQUEST']._serialized_end=2417
-  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2419
-  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2473
-  _globals['_MEMORYUSAGEDATA']._serialized_start=2476
-  _globals['_MEMORYUSAGEDATA']._serialized_end=2618
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2570
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2618
-  _globals['_STATUSRESPONSE']._serialized_start=2621
-  _globals['_STATUSRESPONSE']._serialized_end=2794
-  _globals['_STATUSRESPONSE_STATE']._serialized_start=2727
-  _globals['_STATUSRESPONSE_STATE']._serialized_end=2794
-  _globals['_BACKEND']._serialized_start=2797
-  _globals['_BACKEND']._serialized_end=3425
+  _globals['_MODELOPTIONS']._serialized_end=1834
+  _globals['_RESULT']._serialized_start=1836
+  _globals['_RESULT']._serialized_end=1878
+  _globals['_EMBEDDINGRESULT']._serialized_start=1880
+  _globals['_EMBEDDINGRESULT']._serialized_end=1917
+  _globals['_TRANSCRIPTREQUEST']._serialized_start=1919
+  _globals['_TRANSCRIPTREQUEST']._serialized_end=1986
+  _globals['_TRANSCRIPTRESULT']._serialized_start=1988
+  _globals['_TRANSCRIPTRESULT']._serialized_end=2066
+  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2068
+  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2157
+  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2160
+  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2375
+  _globals['_TTSREQUEST']._serialized_start=2377
+  _globals['_TTSREQUEST']._serialized_end=2431
+  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2433
+  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2487
+  _globals['_MEMORYUSAGEDATA']._serialized_start=2490
+  _globals['_MEMORYUSAGEDATA']._serialized_end=2632
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2584
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2632
+  _globals['_STATUSRESPONSE']._serialized_start=2635
+  _globals['_STATUSRESPONSE']._serialized_end=2808
+  _globals['_STATUSRESPONSE_STATE']._serialized_start=2741
+  _globals['_STATUSRESPONSE_STATE']._serialized_end=2808
+  _globals['_BACKEND']._serialized_start=2811
+  _globals['_BACKEND']._serialized_end=3439
 # @@protoc_insertion_point(module_scope)
diff --git a/backend/python/sentencetransformers/backend_pb2.py b/backend/python/sentencetransformers/backend_pb2.py
index a4a46e04..be6191b1 100644
--- a/backend/python/sentencetransformers/backend_pb2.py
+++ b/backend/python/sentencetransformers/backend_pb2.py
@@ -1,7 +1,6 @@
 # -*- coding: utf-8 -*-
 # Generated by the protocol buffer compiler.  DO NOT EDIT!
 # source: backend.proto
-# Protobuf Python Version: 4.25.0
 """Generated protocol buffer code."""
 from google.protobuf import descriptor as _descriptor
 from google.protobuf import descriptor_pool as _descriptor_pool
@@ -14,16 +13,17 @@ _sym_db = _symbol_database.Default()
 
 
 
-DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xad\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
+DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xbb\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\x12\x0c\n\x04Type\x18\x31 \x01(\t\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
 
 _globals = globals()
 _builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, _globals)
 _builder.BuildTopDescriptorsAndMessages(DESCRIPTOR, 'backend_pb2', _globals)
 if _descriptor._USE_C_DESCRIPTORS == False:
-  _globals['DESCRIPTOR']._options = None
-  _globals['DESCRIPTOR']._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._options = None
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_options = b'8\001'
+
+  DESCRIPTOR._options = None
+  DESCRIPTOR._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._options = None
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._serialized_options = b'8\001'
   _globals['_HEALTHMESSAGE']._serialized_start=26
   _globals['_HEALTHMESSAGE']._serialized_end=41
   _globals['_PREDICTOPTIONS']._serialized_start=44
@@ -31,31 +31,31 @@ if _descriptor._USE_C_DESCRIPTORS == False:
   _globals['_REPLY']._serialized_start=852
   _globals['_REPLY']._serialized_end=876
   _globals['_MODELOPTIONS']._serialized_start=879
-  _globals['_MODELOPTIONS']._serialized_end=1820
-  _globals['_RESULT']._serialized_start=1822
-  _globals['_RESULT']._serialized_end=1864
-  _globals['_EMBEDDINGRESULT']._serialized_start=1866
-  _globals['_EMBEDDINGRESULT']._serialized_end=1903
-  _globals['_TRANSCRIPTREQUEST']._serialized_start=1905
-  _globals['_TRANSCRIPTREQUEST']._serialized_end=1972
-  _globals['_TRANSCRIPTRESULT']._serialized_start=1974
-  _globals['_TRANSCRIPTRESULT']._serialized_end=2052
-  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2054
-  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2143
-  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2146
-  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2361
-  _globals['_TTSREQUEST']._serialized_start=2363
-  _globals['_TTSREQUEST']._serialized_end=2417
-  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2419
-  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2473
-  _globals['_MEMORYUSAGEDATA']._serialized_start=2476
-  _globals['_MEMORYUSAGEDATA']._serialized_end=2618
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2570
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2618
-  _globals['_STATUSRESPONSE']._serialized_start=2621
-  _globals['_STATUSRESPONSE']._serialized_end=2794
-  _globals['_STATUSRESPONSE_STATE']._serialized_start=2727
-  _globals['_STATUSRESPONSE_STATE']._serialized_end=2794
-  _globals['_BACKEND']._serialized_start=2797
-  _globals['_BACKEND']._serialized_end=3425
+  _globals['_MODELOPTIONS']._serialized_end=1834
+  _globals['_RESULT']._serialized_start=1836
+  _globals['_RESULT']._serialized_end=1878
+  _globals['_EMBEDDINGRESULT']._serialized_start=1880
+  _globals['_EMBEDDINGRESULT']._serialized_end=1917
+  _globals['_TRANSCRIPTREQUEST']._serialized_start=1919
+  _globals['_TRANSCRIPTREQUEST']._serialized_end=1986
+  _globals['_TRANSCRIPTRESULT']._serialized_start=1988
+  _globals['_TRANSCRIPTRESULT']._serialized_end=2066
+  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2068
+  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2157
+  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2160
+  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2375
+  _globals['_TTSREQUEST']._serialized_start=2377
+  _globals['_TTSREQUEST']._serialized_end=2431
+  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2433
+  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2487
+  _globals['_MEMORYUSAGEDATA']._serialized_start=2490
+  _globals['_MEMORYUSAGEDATA']._serialized_end=2632
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2584
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2632
+  _globals['_STATUSRESPONSE']._serialized_start=2635
+  _globals['_STATUSRESPONSE']._serialized_end=2808
+  _globals['_STATUSRESPONSE_STATE']._serialized_start=2741
+  _globals['_STATUSRESPONSE_STATE']._serialized_end=2808
+  _globals['_BACKEND']._serialized_start=2811
+  _globals['_BACKEND']._serialized_end=3439
 # @@protoc_insertion_point(module_scope)
diff --git a/backend/python/transformers-musicgen/backend_pb2.py b/backend/python/transformers-musicgen/backend_pb2.py
index a4a46e04..be6191b1 100644
--- a/backend/python/transformers-musicgen/backend_pb2.py
+++ b/backend/python/transformers-musicgen/backend_pb2.py
@@ -1,7 +1,6 @@
 # -*- coding: utf-8 -*-
 # Generated by the protocol buffer compiler.  DO NOT EDIT!
 # source: backend.proto
-# Protobuf Python Version: 4.25.0
 """Generated protocol buffer code."""
 from google.protobuf import descriptor as _descriptor
 from google.protobuf import descriptor_pool as _descriptor_pool
@@ -14,16 +13,17 @@ _sym_db = _symbol_database.Default()
 
 
 
-DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xad\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
+DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xbb\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\x12\x0c\n\x04Type\x18\x31 \x01(\t\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
 
 _globals = globals()
 _builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, _globals)
 _builder.BuildTopDescriptorsAndMessages(DESCRIPTOR, 'backend_pb2', _globals)
 if _descriptor._USE_C_DESCRIPTORS == False:
-  _globals['DESCRIPTOR']._options = None
-  _globals['DESCRIPTOR']._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._options = None
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_options = b'8\001'
+
+  DESCRIPTOR._options = None
+  DESCRIPTOR._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._options = None
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._serialized_options = b'8\001'
   _globals['_HEALTHMESSAGE']._serialized_start=26
   _globals['_HEALTHMESSAGE']._serialized_end=41
   _globals['_PREDICTOPTIONS']._serialized_start=44
@@ -31,31 +31,31 @@ if _descriptor._USE_C_DESCRIPTORS == False:
   _globals['_REPLY']._serialized_start=852
   _globals['_REPLY']._serialized_end=876
   _globals['_MODELOPTIONS']._serialized_start=879
-  _globals['_MODELOPTIONS']._serialized_end=1820
-  _globals['_RESULT']._serialized_start=1822
-  _globals['_RESULT']._serialized_end=1864
-  _globals['_EMBEDDINGRESULT']._serialized_start=1866
-  _globals['_EMBEDDINGRESULT']._serialized_end=1903
-  _globals['_TRANSCRIPTREQUEST']._serialized_start=1905
-  _globals['_TRANSCRIPTREQUEST']._serialized_end=1972
-  _globals['_TRANSCRIPTRESULT']._serialized_start=1974
-  _globals['_TRANSCRIPTRESULT']._serialized_end=2052
-  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2054
-  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2143
-  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2146
-  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2361
-  _globals['_TTSREQUEST']._serialized_start=2363
-  _globals['_TTSREQUEST']._serialized_end=2417
-  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2419
-  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2473
-  _globals['_MEMORYUSAGEDATA']._serialized_start=2476
-  _globals['_MEMORYUSAGEDATA']._serialized_end=2618
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2570
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2618
-  _globals['_STATUSRESPONSE']._serialized_start=2621
-  _globals['_STATUSRESPONSE']._serialized_end=2794
-  _globals['_STATUSRESPONSE_STATE']._serialized_start=2727
-  _globals['_STATUSRESPONSE_STATE']._serialized_end=2794
-  _globals['_BACKEND']._serialized_start=2797
-  _globals['_BACKEND']._serialized_end=3425
+  _globals['_MODELOPTIONS']._serialized_end=1834
+  _globals['_RESULT']._serialized_start=1836
+  _globals['_RESULT']._serialized_end=1878
+  _globals['_EMBEDDINGRESULT']._serialized_start=1880
+  _globals['_EMBEDDINGRESULT']._serialized_end=1917
+  _globals['_TRANSCRIPTREQUEST']._serialized_start=1919
+  _globals['_TRANSCRIPTREQUEST']._serialized_end=1986
+  _globals['_TRANSCRIPTRESULT']._serialized_start=1988
+  _globals['_TRANSCRIPTRESULT']._serialized_end=2066
+  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2068
+  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2157
+  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2160
+  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2375
+  _globals['_TTSREQUEST']._serialized_start=2377
+  _globals['_TTSREQUEST']._serialized_end=2431
+  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2433
+  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2487
+  _globals['_MEMORYUSAGEDATA']._serialized_start=2490
+  _globals['_MEMORYUSAGEDATA']._serialized_end=2632
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2584
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2632
+  _globals['_STATUSRESPONSE']._serialized_start=2635
+  _globals['_STATUSRESPONSE']._serialized_end=2808
+  _globals['_STATUSRESPONSE_STATE']._serialized_start=2741
+  _globals['_STATUSRESPONSE_STATE']._serialized_end=2808
+  _globals['_BACKEND']._serialized_start=2811
+  _globals['_BACKEND']._serialized_end=3439
 # @@protoc_insertion_point(module_scope)
diff --git a/backend/python/transformers/backend_pb2.py b/backend/python/transformers/backend_pb2.py
index a4a46e04..be6191b1 100644
--- a/backend/python/transformers/backend_pb2.py
+++ b/backend/python/transformers/backend_pb2.py
@@ -1,7 +1,6 @@
 # -*- coding: utf-8 -*-
 # Generated by the protocol buffer compiler.  DO NOT EDIT!
 # source: backend.proto
-# Protobuf Python Version: 4.25.0
 """Generated protocol buffer code."""
 from google.protobuf import descriptor as _descriptor
 from google.protobuf import descriptor_pool as _descriptor_pool
@@ -14,16 +13,17 @@ _sym_db = _symbol_database.Default()
 
 
 
-DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xad\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
+DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xbb\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\x12\x0c\n\x04Type\x18\x31 \x01(\t\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
 
 _globals = globals()
 _builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, _globals)
 _builder.BuildTopDescriptorsAndMessages(DESCRIPTOR, 'backend_pb2', _globals)
 if _descriptor._USE_C_DESCRIPTORS == False:
-  _globals['DESCRIPTOR']._options = None
-  _globals['DESCRIPTOR']._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._options = None
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_options = b'8\001'
+
+  DESCRIPTOR._options = None
+  DESCRIPTOR._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._options = None
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._serialized_options = b'8\001'
   _globals['_HEALTHMESSAGE']._serialized_start=26
   _globals['_HEALTHMESSAGE']._serialized_end=41
   _globals['_PREDICTOPTIONS']._serialized_start=44
@@ -31,31 +31,31 @@ if _descriptor._USE_C_DESCRIPTORS == False:
   _globals['_REPLY']._serialized_start=852
   _globals['_REPLY']._serialized_end=876
   _globals['_MODELOPTIONS']._serialized_start=879
-  _globals['_MODELOPTIONS']._serialized_end=1820
-  _globals['_RESULT']._serialized_start=1822
-  _globals['_RESULT']._serialized_end=1864
-  _globals['_EMBEDDINGRESULT']._serialized_start=1866
-  _globals['_EMBEDDINGRESULT']._serialized_end=1903
-  _globals['_TRANSCRIPTREQUEST']._serialized_start=1905
-  _globals['_TRANSCRIPTREQUEST']._serialized_end=1972
-  _globals['_TRANSCRIPTRESULT']._serialized_start=1974
-  _globals['_TRANSCRIPTRESULT']._serialized_end=2052
-  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2054
-  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2143
-  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2146
-  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2361
-  _globals['_TTSREQUEST']._serialized_start=2363
-  _globals['_TTSREQUEST']._serialized_end=2417
-  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2419
-  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2473
-  _globals['_MEMORYUSAGEDATA']._serialized_start=2476
-  _globals['_MEMORYUSAGEDATA']._serialized_end=2618
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2570
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2618
-  _globals['_STATUSRESPONSE']._serialized_start=2621
-  _globals['_STATUSRESPONSE']._serialized_end=2794
-  _globals['_STATUSRESPONSE_STATE']._serialized_start=2727
-  _globals['_STATUSRESPONSE_STATE']._serialized_end=2794
-  _globals['_BACKEND']._serialized_start=2797
-  _globals['_BACKEND']._serialized_end=3425
+  _globals['_MODELOPTIONS']._serialized_end=1834
+  _globals['_RESULT']._serialized_start=1836
+  _globals['_RESULT']._serialized_end=1878
+  _globals['_EMBEDDINGRESULT']._serialized_start=1880
+  _globals['_EMBEDDINGRESULT']._serialized_end=1917
+  _globals['_TRANSCRIPTREQUEST']._serialized_start=1919
+  _globals['_TRANSCRIPTREQUEST']._serialized_end=1986
+  _globals['_TRANSCRIPTRESULT']._serialized_start=1988
+  _globals['_TRANSCRIPTRESULT']._serialized_end=2066
+  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2068
+  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2157
+  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2160
+  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2375
+  _globals['_TTSREQUEST']._serialized_start=2377
+  _globals['_TTSREQUEST']._serialized_end=2431
+  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2433
+  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2487
+  _globals['_MEMORYUSAGEDATA']._serialized_start=2490
+  _globals['_MEMORYUSAGEDATA']._serialized_end=2632
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2584
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2632
+  _globals['_STATUSRESPONSE']._serialized_start=2635
+  _globals['_STATUSRESPONSE']._serialized_end=2808
+  _globals['_STATUSRESPONSE_STATE']._serialized_start=2741
+  _globals['_STATUSRESPONSE_STATE']._serialized_end=2808
+  _globals['_BACKEND']._serialized_start=2811
+  _globals['_BACKEND']._serialized_end=3439
 # @@protoc_insertion_point(module_scope)
diff --git a/backend/python/transformers/transformers_server.py b/backend/python/transformers/transformers_server.py
index e6c06bab..1b177057 100755
--- a/backend/python/transformers/transformers_server.py
+++ b/backend/python/transformers/transformers_server.py
@@ -16,7 +16,7 @@ import backend_pb2_grpc
 import grpc
 import torch
 import torch.cuda
-from transformers import AutoTokenizer, AutoModel
+from transformers import AutoTokenizer, AutoModel, AutoModelForCausalLM, set_seed
 
 _ONE_DAY_IN_SECONDS = 60 * 60 * 24
 
@@ -68,12 +68,19 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
         """
         model_name = request.Model
         try:
-            self.model = AutoModel.from_pretrained(model_name, trust_remote_code=True) # trust_remote_code is needed to use the encode method with embeddings models like jinai-v2
+            if request.Type == "AutoModelForCausalLM":
+                self.model = AutoModelForCausalLM.from_pretrained(model_name, trust_remote_code=True)
+            else:
+                self.model = AutoModel.from_pretrained(model_name, trust_remote_code=True)
+
             self.tokenizer = AutoTokenizer.from_pretrained(model_name)
+            self.CUDA = False
+
             if request.CUDA or torch.cuda.is_available():
                 try:
                     print("Loading model", model_name, "to CUDA.", file=sys.stderr)
                     self.model = self.model.to("cuda")
+                    self.CUDA = True
                 except Exception as err:
                     print("Not using CUDA:", err, file=sys.stderr)
         except Exception as err:
@@ -94,6 +101,7 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
             An EmbeddingResult object that contains the calculated embeddings.
         """
 
+        set_seed(request.Seed)
         # Tokenize input
         max_length = 512
         if request.Tokens != 0:
@@ -120,6 +128,7 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
         Returns:
             backend_pb2.Reply: The predict result.
         """
+        set_seed(request.Seed)
         if request.TopP == 0:
             request.TopP = 0.9
 
@@ -127,10 +136,13 @@ class BackendServicer(backend_pb2_grpc.BackendServicer):
         if request.Tokens > 0:
             max_tokens = request.Tokens
 
-        inputs = self.tokenizer.tokenizer(request.Prompt, return_tensors="pt").input_ids
-        outputs = self.model.generate(inputs,max_tokens=max_tokens, temperature=request.Temperature, top_p=request.TopP)
+        inputs = self.tokenizer(request.Prompt, return_tensors="pt").input_ids
+        if self.CUDA:
+            inputs = inputs.to("cuda")
 
-        generated_text = self.tokenizer.tokenizer.batch_decode(outputs, skip_special_tokens=True)[0]
+        outputs = self.model.generate(inputs,max_new_tokens=max_tokens, temperature=request.Temperature, top_p=request.TopP)
+
+        generated_text = self.tokenizer.batch_decode(outputs, skip_special_tokens=True)[0]
         # Remove prompt from response if present
         if request.Prompt in generated_text:
             generated_text = generated_text.replace(request.Prompt, "")
diff --git a/backend/python/vall-e-x/backend_pb2.py b/backend/python/vall-e-x/backend_pb2.py
index a4a46e04..be6191b1 100644
--- a/backend/python/vall-e-x/backend_pb2.py
+++ b/backend/python/vall-e-x/backend_pb2.py
@@ -1,7 +1,6 @@
 # -*- coding: utf-8 -*-
 # Generated by the protocol buffer compiler.  DO NOT EDIT!
 # source: backend.proto
-# Protobuf Python Version: 4.25.0
 """Generated protocol buffer code."""
 from google.protobuf import descriptor as _descriptor
 from google.protobuf import descriptor_pool as _descriptor_pool
@@ -14,16 +13,17 @@ _sym_db = _symbol_database.Default()
 
 
 
-DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xad\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
+DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xbb\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\x12\x0c\n\x04Type\x18\x31 \x01(\t\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
 
 _globals = globals()
 _builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, _globals)
 _builder.BuildTopDescriptorsAndMessages(DESCRIPTOR, 'backend_pb2', _globals)
 if _descriptor._USE_C_DESCRIPTORS == False:
-  _globals['DESCRIPTOR']._options = None
-  _globals['DESCRIPTOR']._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._options = None
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_options = b'8\001'
+
+  DESCRIPTOR._options = None
+  DESCRIPTOR._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._options = None
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._serialized_options = b'8\001'
   _globals['_HEALTHMESSAGE']._serialized_start=26
   _globals['_HEALTHMESSAGE']._serialized_end=41
   _globals['_PREDICTOPTIONS']._serialized_start=44
@@ -31,31 +31,31 @@ if _descriptor._USE_C_DESCRIPTORS == False:
   _globals['_REPLY']._serialized_start=852
   _globals['_REPLY']._serialized_end=876
   _globals['_MODELOPTIONS']._serialized_start=879
-  _globals['_MODELOPTIONS']._serialized_end=1820
-  _globals['_RESULT']._serialized_start=1822
-  _globals['_RESULT']._serialized_end=1864
-  _globals['_EMBEDDINGRESULT']._serialized_start=1866
-  _globals['_EMBEDDINGRESULT']._serialized_end=1903
-  _globals['_TRANSCRIPTREQUEST']._serialized_start=1905
-  _globals['_TRANSCRIPTREQUEST']._serialized_end=1972
-  _globals['_TRANSCRIPTRESULT']._serialized_start=1974
-  _globals['_TRANSCRIPTRESULT']._serialized_end=2052
-  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2054
-  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2143
-  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2146
-  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2361
-  _globals['_TTSREQUEST']._serialized_start=2363
-  _globals['_TTSREQUEST']._serialized_end=2417
-  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2419
-  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2473
-  _globals['_MEMORYUSAGEDATA']._serialized_start=2476
-  _globals['_MEMORYUSAGEDATA']._serialized_end=2618
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2570
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2618
-  _globals['_STATUSRESPONSE']._serialized_start=2621
-  _globals['_STATUSRESPONSE']._serialized_end=2794
-  _globals['_STATUSRESPONSE_STATE']._serialized_start=2727
-  _globals['_STATUSRESPONSE_STATE']._serialized_end=2794
-  _globals['_BACKEND']._serialized_start=2797
-  _globals['_BACKEND']._serialized_end=3425
+  _globals['_MODELOPTIONS']._serialized_end=1834
+  _globals['_RESULT']._serialized_start=1836
+  _globals['_RESULT']._serialized_end=1878
+  _globals['_EMBEDDINGRESULT']._serialized_start=1880
+  _globals['_EMBEDDINGRESULT']._serialized_end=1917
+  _globals['_TRANSCRIPTREQUEST']._serialized_start=1919
+  _globals['_TRANSCRIPTREQUEST']._serialized_end=1986
+  _globals['_TRANSCRIPTRESULT']._serialized_start=1988
+  _globals['_TRANSCRIPTRESULT']._serialized_end=2066
+  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2068
+  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2157
+  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2160
+  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2375
+  _globals['_TTSREQUEST']._serialized_start=2377
+  _globals['_TTSREQUEST']._serialized_end=2431
+  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2433
+  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2487
+  _globals['_MEMORYUSAGEDATA']._serialized_start=2490
+  _globals['_MEMORYUSAGEDATA']._serialized_end=2632
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2584
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2632
+  _globals['_STATUSRESPONSE']._serialized_start=2635
+  _globals['_STATUSRESPONSE']._serialized_end=2808
+  _globals['_STATUSRESPONSE_STATE']._serialized_start=2741
+  _globals['_STATUSRESPONSE_STATE']._serialized_end=2808
+  _globals['_BACKEND']._serialized_start=2811
+  _globals['_BACKEND']._serialized_end=3439
 # @@protoc_insertion_point(module_scope)
diff --git a/backend/python/vllm/backend_pb2.py b/backend/python/vllm/backend_pb2.py
index a4a46e04..be6191b1 100644
--- a/backend/python/vllm/backend_pb2.py
+++ b/backend/python/vllm/backend_pb2.py
@@ -1,7 +1,6 @@
 # -*- coding: utf-8 -*-
 # Generated by the protocol buffer compiler.  DO NOT EDIT!
 # source: backend.proto
-# Protobuf Python Version: 4.25.0
 """Generated protocol buffer code."""
 from google.protobuf import descriptor as _descriptor
 from google.protobuf import descriptor_pool as _descriptor_pool
@@ -14,16 +13,17 @@ _sym_db = _symbol_database.Default()
 
 
 
-DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xad\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
+DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\rbackend.proto\x12\x07\x62\x61\x63kend\"\x0f\n\rHealthMessage\"\xa6\x06\n\x0ePredictOptions\x12\x0e\n\x06Prompt\x18\x01 \x01(\t\x12\x0c\n\x04Seed\x18\x02 \x01(\x05\x12\x0f\n\x07Threads\x18\x03 \x01(\x05\x12\x0e\n\x06Tokens\x18\x04 \x01(\x05\x12\x0c\n\x04TopK\x18\x05 \x01(\x05\x12\x0e\n\x06Repeat\x18\x06 \x01(\x05\x12\r\n\x05\x42\x61tch\x18\x07 \x01(\x05\x12\r\n\x05NKeep\x18\x08 \x01(\x05\x12\x13\n\x0bTemperature\x18\t \x01(\x02\x12\x0f\n\x07Penalty\x18\n \x01(\x02\x12\r\n\x05\x46\x31\x36KV\x18\x0b \x01(\x08\x12\x11\n\tDebugMode\x18\x0c \x01(\x08\x12\x13\n\x0bStopPrompts\x18\r \x03(\t\x12\x11\n\tIgnoreEOS\x18\x0e \x01(\x08\x12\x19\n\x11TailFreeSamplingZ\x18\x0f \x01(\x02\x12\x10\n\x08TypicalP\x18\x10 \x01(\x02\x12\x18\n\x10\x46requencyPenalty\x18\x11 \x01(\x02\x12\x17\n\x0fPresencePenalty\x18\x12 \x01(\x02\x12\x10\n\x08Mirostat\x18\x13 \x01(\x05\x12\x13\n\x0bMirostatETA\x18\x14 \x01(\x02\x12\x13\n\x0bMirostatTAU\x18\x15 \x01(\x02\x12\x12\n\nPenalizeNL\x18\x16 \x01(\x08\x12\x11\n\tLogitBias\x18\x17 \x01(\t\x12\r\n\x05MLock\x18\x19 \x01(\x08\x12\x0c\n\x04MMap\x18\x1a \x01(\x08\x12\x16\n\x0ePromptCacheAll\x18\x1b \x01(\x08\x12\x15\n\rPromptCacheRO\x18\x1c \x01(\x08\x12\x0f\n\x07Grammar\x18\x1d \x01(\t\x12\x0f\n\x07MainGPU\x18\x1e \x01(\t\x12\x13\n\x0bTensorSplit\x18\x1f \x01(\t\x12\x0c\n\x04TopP\x18  \x01(\x02\x12\x17\n\x0fPromptCachePath\x18! \x01(\t\x12\r\n\x05\x44\x65\x62ug\x18\" \x01(\x08\x12\x17\n\x0f\x45mbeddingTokens\x18# \x03(\x05\x12\x12\n\nEmbeddings\x18$ \x01(\t\x12\x14\n\x0cRopeFreqBase\x18% \x01(\x02\x12\x15\n\rRopeFreqScale\x18& \x01(\x02\x12\x1b\n\x13NegativePromptScale\x18\' \x01(\x02\x12\x16\n\x0eNegativePrompt\x18( \x01(\t\x12\x0e\n\x06NDraft\x18) \x01(\x05\x12\x0e\n\x06Images\x18* \x03(\t\"\x18\n\x05Reply\x12\x0f\n\x07message\x18\x01 \x01(\x0c\"\xbb\x07\n\x0cModelOptions\x12\r\n\x05Model\x18\x01 \x01(\t\x12\x13\n\x0b\x43ontextSize\x18\x02 \x01(\x05\x12\x0c\n\x04Seed\x18\x03 \x01(\x05\x12\x0e\n\x06NBatch\x18\x04 \x01(\x05\x12\x11\n\tF16Memory\x18\x05 \x01(\x08\x12\r\n\x05MLock\x18\x06 \x01(\x08\x12\x0c\n\x04MMap\x18\x07 \x01(\x08\x12\x11\n\tVocabOnly\x18\x08 \x01(\x08\x12\x0f\n\x07LowVRAM\x18\t \x01(\x08\x12\x12\n\nEmbeddings\x18\n \x01(\x08\x12\x0c\n\x04NUMA\x18\x0b \x01(\x08\x12\x12\n\nNGPULayers\x18\x0c \x01(\x05\x12\x0f\n\x07MainGPU\x18\r \x01(\t\x12\x13\n\x0bTensorSplit\x18\x0e \x01(\t\x12\x0f\n\x07Threads\x18\x0f \x01(\x05\x12\x19\n\x11LibrarySearchPath\x18\x10 \x01(\t\x12\x14\n\x0cRopeFreqBase\x18\x11 \x01(\x02\x12\x15\n\rRopeFreqScale\x18\x12 \x01(\x02\x12\x12\n\nRMSNormEps\x18\x13 \x01(\x02\x12\x0c\n\x04NGQA\x18\x14 \x01(\x05\x12\x11\n\tModelFile\x18\x15 \x01(\t\x12\x0e\n\x06\x44\x65vice\x18\x16 \x01(\t\x12\x11\n\tUseTriton\x18\x17 \x01(\x08\x12\x15\n\rModelBaseName\x18\x18 \x01(\t\x12\x18\n\x10UseFastTokenizer\x18\x19 \x01(\x08\x12\x14\n\x0cPipelineType\x18\x1a \x01(\t\x12\x15\n\rSchedulerType\x18\x1b \x01(\t\x12\x0c\n\x04\x43UDA\x18\x1c \x01(\x08\x12\x10\n\x08\x43\x46GScale\x18\x1d \x01(\x02\x12\x0f\n\x07IMG2IMG\x18\x1e \x01(\x08\x12\x11\n\tCLIPModel\x18\x1f \x01(\t\x12\x15\n\rCLIPSubfolder\x18  \x01(\t\x12\x10\n\x08\x43LIPSkip\x18! \x01(\x05\x12\x12\n\nControlNet\x18\x30 \x01(\t\x12\x11\n\tTokenizer\x18\" \x01(\t\x12\x10\n\x08LoraBase\x18# \x01(\t\x12\x13\n\x0bLoraAdapter\x18$ \x01(\t\x12\x11\n\tLoraScale\x18* \x01(\x02\x12\x11\n\tNoMulMatQ\x18% \x01(\x08\x12\x12\n\nDraftModel\x18\' \x01(\t\x12\x11\n\tAudioPath\x18& \x01(\t\x12\x14\n\x0cQuantization\x18( \x01(\t\x12\x0e\n\x06MMProj\x18) \x01(\t\x12\x13\n\x0bRopeScaling\x18+ \x01(\t\x12\x15\n\rYarnExtFactor\x18, \x01(\x02\x12\x16\n\x0eYarnAttnFactor\x18- \x01(\x02\x12\x14\n\x0cYarnBetaFast\x18. \x01(\x02\x12\x14\n\x0cYarnBetaSlow\x18/ \x01(\x02\x12\x0c\n\x04Type\x18\x31 \x01(\t\"*\n\x06Result\x12\x0f\n\x07message\x18\x01 \x01(\t\x12\x0f\n\x07success\x18\x02 \x01(\x08\"%\n\x0f\x45mbeddingResult\x12\x12\n\nembeddings\x18\x01 \x03(\x02\"C\n\x11TranscriptRequest\x12\x0b\n\x03\x64st\x18\x02 \x01(\t\x12\x10\n\x08language\x18\x03 \x01(\t\x12\x0f\n\x07threads\x18\x04 \x01(\r\"N\n\x10TranscriptResult\x12,\n\x08segments\x18\x01 \x03(\x0b\x32\x1a.backend.TranscriptSegment\x12\x0c\n\x04text\x18\x02 \x01(\t\"Y\n\x11TranscriptSegment\x12\n\n\x02id\x18\x01 \x01(\x05\x12\r\n\x05start\x18\x02 \x01(\x03\x12\x0b\n\x03\x65nd\x18\x03 \x01(\x03\x12\x0c\n\x04text\x18\x04 \x01(\t\x12\x0e\n\x06tokens\x18\x05 \x03(\x05\"\xd7\x01\n\x14GenerateImageRequest\x12\x0e\n\x06height\x18\x01 \x01(\x05\x12\r\n\x05width\x18\x02 \x01(\x05\x12\x0c\n\x04mode\x18\x03 \x01(\x05\x12\x0c\n\x04step\x18\x04 \x01(\x05\x12\x0c\n\x04seed\x18\x05 \x01(\x05\x12\x17\n\x0fpositive_prompt\x18\x06 \x01(\t\x12\x17\n\x0fnegative_prompt\x18\x07 \x01(\t\x12\x0b\n\x03\x64st\x18\x08 \x01(\t\x12\x0b\n\x03src\x18\t \x01(\t\x12\x18\n\x10\x45nableParameters\x18\n \x01(\t\x12\x10\n\x08\x43LIPSkip\x18\x0b \x01(\x05\"6\n\nTTSRequest\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\r\n\x05model\x18\x02 \x01(\t\x12\x0b\n\x03\x64st\x18\x03 \x01(\t\"6\n\x14TokenizationResponse\x12\x0e\n\x06length\x18\x01 \x01(\x05\x12\x0e\n\x06tokens\x18\x02 \x03(\x05\"\x8e\x01\n\x0fMemoryUsageData\x12\r\n\x05total\x18\x01 \x01(\x04\x12:\n\tbreakdown\x18\x02 \x03(\x0b\x32\'.backend.MemoryUsageData.BreakdownEntry\x1a\x30\n\x0e\x42reakdownEntry\x12\x0b\n\x03key\x18\x01 \x01(\t\x12\r\n\x05value\x18\x02 \x01(\x04:\x02\x38\x01\"\xad\x01\n\x0eStatusResponse\x12,\n\x05state\x18\x01 \x01(\x0e\x32\x1d.backend.StatusResponse.State\x12(\n\x06memory\x18\x02 \x01(\x0b\x32\x18.backend.MemoryUsageData\"C\n\x05State\x12\x11\n\rUNINITIALIZED\x10\x00\x12\x08\n\x04\x42USY\x10\x01\x12\t\n\x05READY\x10\x02\x12\x12\n\x05\x45RROR\x10\xff\xff\xff\xff\xff\xff\xff\xff\xff\x01\x32\xf4\x04\n\x07\x42\x61\x63kend\x12\x32\n\x06Health\x12\x16.backend.HealthMessage\x1a\x0e.backend.Reply\"\x00\x12\x34\n\x07Predict\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x12\x35\n\tLoadModel\x12\x15.backend.ModelOptions\x1a\x0f.backend.Result\"\x00\x12<\n\rPredictStream\x12\x17.backend.PredictOptions\x1a\x0e.backend.Reply\"\x00\x30\x01\x12@\n\tEmbedding\x12\x17.backend.PredictOptions\x1a\x18.backend.EmbeddingResult\"\x00\x12\x41\n\rGenerateImage\x12\x1d.backend.GenerateImageRequest\x1a\x0f.backend.Result\"\x00\x12M\n\x12\x41udioTranscription\x12\x1a.backend.TranscriptRequest\x1a\x19.backend.TranscriptResult\"\x00\x12-\n\x03TTS\x12\x13.backend.TTSRequest\x1a\x0f.backend.Result\"\x00\x12J\n\x0eTokenizeString\x12\x17.backend.PredictOptions\x1a\x1d.backend.TokenizationResponse\"\x00\x12;\n\x06Status\x12\x16.backend.HealthMessage\x1a\x17.backend.StatusResponse\"\x00\x42Z\n\x19io.skynet.localai.backendB\x0eLocalAIBackendP\x01Z+github.com/go-skynet/LocalAI/pkg/grpc/protob\x06proto3')
 
 _globals = globals()
 _builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, _globals)
 _builder.BuildTopDescriptorsAndMessages(DESCRIPTOR, 'backend_pb2', _globals)
 if _descriptor._USE_C_DESCRIPTORS == False:
-  _globals['DESCRIPTOR']._options = None
-  _globals['DESCRIPTOR']._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._options = None
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_options = b'8\001'
+
+  DESCRIPTOR._options = None
+  DESCRIPTOR._serialized_options = b'\n\031io.skynet.localai.backendB\016LocalAIBackendP\001Z+github.com/go-skynet/LocalAI/pkg/grpc/proto'
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._options = None
+  _MEMORYUSAGEDATA_BREAKDOWNENTRY._serialized_options = b'8\001'
   _globals['_HEALTHMESSAGE']._serialized_start=26
   _globals['_HEALTHMESSAGE']._serialized_end=41
   _globals['_PREDICTOPTIONS']._serialized_start=44
@@ -31,31 +31,31 @@ if _descriptor._USE_C_DESCRIPTORS == False:
   _globals['_REPLY']._serialized_start=852
   _globals['_REPLY']._serialized_end=876
   _globals['_MODELOPTIONS']._serialized_start=879
-  _globals['_MODELOPTIONS']._serialized_end=1820
-  _globals['_RESULT']._serialized_start=1822
-  _globals['_RESULT']._serialized_end=1864
-  _globals['_EMBEDDINGRESULT']._serialized_start=1866
-  _globals['_EMBEDDINGRESULT']._serialized_end=1903
-  _globals['_TRANSCRIPTREQUEST']._serialized_start=1905
-  _globals['_TRANSCRIPTREQUEST']._serialized_end=1972
-  _globals['_TRANSCRIPTRESULT']._serialized_start=1974
-  _globals['_TRANSCRIPTRESULT']._serialized_end=2052
-  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2054
-  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2143
-  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2146
-  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2361
-  _globals['_TTSREQUEST']._serialized_start=2363
-  _globals['_TTSREQUEST']._serialized_end=2417
-  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2419
-  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2473
-  _globals['_MEMORYUSAGEDATA']._serialized_start=2476
-  _globals['_MEMORYUSAGEDATA']._serialized_end=2618
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2570
-  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2618
-  _globals['_STATUSRESPONSE']._serialized_start=2621
-  _globals['_STATUSRESPONSE']._serialized_end=2794
-  _globals['_STATUSRESPONSE_STATE']._serialized_start=2727
-  _globals['_STATUSRESPONSE_STATE']._serialized_end=2794
-  _globals['_BACKEND']._serialized_start=2797
-  _globals['_BACKEND']._serialized_end=3425
+  _globals['_MODELOPTIONS']._serialized_end=1834
+  _globals['_RESULT']._serialized_start=1836
+  _globals['_RESULT']._serialized_end=1878
+  _globals['_EMBEDDINGRESULT']._serialized_start=1880
+  _globals['_EMBEDDINGRESULT']._serialized_end=1917
+  _globals['_TRANSCRIPTREQUEST']._serialized_start=1919
+  _globals['_TRANSCRIPTREQUEST']._serialized_end=1986
+  _globals['_TRANSCRIPTRESULT']._serialized_start=1988
+  _globals['_TRANSCRIPTRESULT']._serialized_end=2066
+  _globals['_TRANSCRIPTSEGMENT']._serialized_start=2068
+  _globals['_TRANSCRIPTSEGMENT']._serialized_end=2157
+  _globals['_GENERATEIMAGEREQUEST']._serialized_start=2160
+  _globals['_GENERATEIMAGEREQUEST']._serialized_end=2375
+  _globals['_TTSREQUEST']._serialized_start=2377
+  _globals['_TTSREQUEST']._serialized_end=2431
+  _globals['_TOKENIZATIONRESPONSE']._serialized_start=2433
+  _globals['_TOKENIZATIONRESPONSE']._serialized_end=2487
+  _globals['_MEMORYUSAGEDATA']._serialized_start=2490
+  _globals['_MEMORYUSAGEDATA']._serialized_end=2632
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_start=2584
+  _globals['_MEMORYUSAGEDATA_BREAKDOWNENTRY']._serialized_end=2632
+  _globals['_STATUSRESPONSE']._serialized_start=2635
+  _globals['_STATUSRESPONSE']._serialized_end=2808
+  _globals['_STATUSRESPONSE_STATE']._serialized_start=2741
+  _globals['_STATUSRESPONSE_STATE']._serialized_end=2808
+  _globals['_BACKEND']._serialized_start=2811
+  _globals['_BACKEND']._serialized_end=3439
 # @@protoc_insertion_point(module_scope)
diff --git a/docs/content/docs/getting-started/customize-model.md b/docs/content/docs/getting-started/customize-model.md
index 0984d6ce..0d1cf21a 100644
--- a/docs/content/docs/getting-started/customize-model.md
+++ b/docs/content/docs/getting-started/customize-model.md
@@ -26,6 +26,8 @@ Here's an example to initiate the **phi-2** model:
 docker run -p 8080:8080 localai/localai:{{< version >}}-ffmpeg-core https://gist.githubusercontent.com/mudler/ad601a0488b497b69ec549150d9edd18/raw/a8a8869ef1bb7e3830bf5c0bae29a0cce991ff8d/phi-2.yaml
 ```
 
+You can also check all the embedded models configurations [here](https://github.com/mudler/LocalAI/tree/master/embedded/models).
+
 {{% alert icon="" %}}
 The model configurations used in the quickstart are accessible here: [https://github.com/mudler/LocalAI/tree/master/embedded/models](https://github.com/mudler/LocalAI/tree/master/embedded/models). Contributions are welcome; please feel free to submit a Pull Request.
 
diff --git a/docs/content/docs/getting-started/quickstart.md b/docs/content/docs/getting-started/quickstart.md
index 89833d82..23d7d70f 100644
--- a/docs/content/docs/getting-started/quickstart.md
+++ b/docs/content/docs/getting-started/quickstart.md
@@ -40,7 +40,7 @@ There are different categories of models: [LLMs]({{%relref "docs/features/text-g
 
 {{% alert icon="💡" %}}
 
-To customize the models, see [Model customization]({{%relref "docs/getting-started/customize-model" %}}). For more model configurations, visit the [Examples Section](https://github.com/mudler/LocalAI/tree/master/examples/configurations).
+To customize the models, see [Model customization]({{%relref "docs/getting-started/customize-model" %}}). For more model configurations, visit the [Examples Section](https://github.com/mudler/LocalAI/tree/master/examples/configurations) and the configurations for the models below is available [here](https://github.com/mudler/LocalAI/tree/master/embedded/models).
 {{% /alert %}}
 
 {{< tabs tabTotal="3" >}}
@@ -64,6 +64,10 @@ To customize the models, see [Model customization]({{%relref "docs/getting-start
 | [tinyllama-chat](https://huggingface.co/TheBloke/TinyLlama-1.1B-Chat-v0.3-GGUF) [original model](https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v0.3) | [LLM]({{%relref "docs/features/text-generation" %}}) | ```docker run -ti -p 8080:8080 localai/localai:{{< version >}}-ffmpeg-core tinyllama-chat``` |
 | [dolphin-2.5-mixtral-8x7b](https://huggingface.co/TheBloke/dolphin-2.5-mixtral-8x7b-GGUF) | [LLM]({{%relref "docs/features/text-generation" %}}) | ```docker run -ti -p 8080:8080 localai/localai:{{< version >}}-ffmpeg-core dolphin-2.5-mixtral-8x7b``` |
 | 🐍 [mamba](https://github.com/state-spaces/mamba) | [LLM]({{%relref "docs/features/text-generation" %}}) | GPU-only |
+| animagine-xl | [Text to Image]({{%relref "docs/features/image-generation" %}}) | GPU-only |
+| transformers-tinyllama | [LLM]({{%relref "docs/features/text-generation" %}}) | GPU-only |
+| [codellama-7b](https://huggingface.co/codellama/CodeLlama-7b-hf) (with transformers) | [LLM]({{%relref "docs/features/text-generation" %}}) | GPU-only |
+| [codellama-7b-gguf](https://huggingface.co/TheBloke/CodeLlama-7B-GGUF) (with llama.cpp) | [LLM]({{%relref "docs/features/text-generation" %}}) | ```docker run -ti -p 8080:8080 localai/localai:{{< version >}}-ffmpeg-core codellama-7b-gguf``` |
 {{% /tab %}}
 {{% tab tabName="GPU (CUDA 11)" %}}
 
@@ -86,7 +90,10 @@ To customize the models, see [Model customization]({{%relref "docs/getting-start
 | [tinyllama-chat](https://huggingface.co/TheBloke/TinyLlama-1.1B-Chat-v0.3-GGUF) [original model](https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v0.3) | [LLM]({{%relref "docs/features/text-generation" %}}) | ```docker run -ti -p 8080:8080 --gpus all localai/localai:{{< version >}}-cublas-cuda11-core tinyllama-chat``` |
 | [dolphin-2.5-mixtral-8x7b](https://huggingface.co/TheBloke/dolphin-2.5-mixtral-8x7b-GGUF) | [LLM]({{%relref "docs/features/text-generation" %}}) | ```docker run -ti -p 8080:8080 --gpus all localai/localai:{{< version >}}-cublas-cuda11-core dolphin-2.5-mixtral-8x7b``` |
 | 🐍 [mamba](https://github.com/state-spaces/mamba) | [LLM]({{%relref "docs/features/text-generation" %}}) | ```docker run -ti -p 8080:8080 --gpus all localai/localai:{{< version >}}-cublas-cuda11 mamba-chat``` |
-
+| animagine-xl | [Text to Image]({{%relref "docs/features/image-generation" %}}) |  ```docker run -ti -p 8080:8080 -e COMPEL=0 --gpus all localai/localai:{{< version >}}-cublas-cuda11 animagine-xl``` |
+| transformers-tinyllama | [LLM]({{%relref "docs/features/text-generation" %}}) | ```docker run -ti -p 8080:8080 --gpus all localai/localai:{{< version >}}-cublas-cuda11 transformers-tinyllama``` |
+| [codellama-7b](https://huggingface.co/codellama/CodeLlama-7b-hf) | [LLM]({{%relref "docs/features/text-generation" %}})  | ```docker run -ti -p 8080:8080 --gpus all localai/localai:{{< version >}}-cublas-cuda11 codellama-7b``` |
+| [codellama-7b-gguf](https://huggingface.co/TheBloke/CodeLlama-7B-GGUF) | [LLM]({{%relref "docs/features/text-generation" %}})  | ```docker run -ti -p 8080:8080 --gpus all localai/localai:{{< version >}}-cublas-cuda11-core codellama-7b-gguf``` |
 {{% /tab %}}
 
 
@@ -110,6 +117,10 @@ To customize the models, see [Model customization]({{%relref "docs/getting-start
 | [tinyllama-chat](https://huggingface.co/TheBloke/TinyLlama-1.1B-Chat-v0.3-GGUF) [original model](https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v0.3) | [LLM]({{%relref "docs/features/text-generation" %}}) | ```docker run -ti -p 8080:8080 --gpus all localai/localai:{{< version >}}-cublas-cuda12-core tinyllama-chat``` |
 | [dolphin-2.5-mixtral-8x7b](https://huggingface.co/TheBloke/dolphin-2.5-mixtral-8x7b-GGUF) | [LLM]({{%relref "docs/features/text-generation" %}}) | ```docker run -ti -p 8080:8080 --gpus all localai/localai:{{< version >}}-cublas-cuda12-core dolphin-2.5-mixtral-8x7b``` |
 | 🐍 [mamba](https://github.com/state-spaces/mamba) | [LLM]({{%relref "docs/features/text-generation" %}}) | ```docker run -ti -p 8080:8080 --gpus all localai/localai:{{< version >}}-cublas-cuda12 mamba-chat``` |
+| animagine-xl | [Text to Image]({{%relref "docs/features/image-generation" %}}) | GPU-only | ```docker run -ti -p 8080:8080 -e COMPEL=0 --gpus all localai/localai:{{< version >}}-cublas-cuda12 animagine-xl``` |
+| transformers-tinyllama | [LLM]({{%relref "docs/features/text-generation" %}}) | GPU-only | ```docker run -ti -p 8080:8080 --gpus all localai/localai:{{< version >}}-cublas-cuda12 transformers-tinyllama``` |
+| [codellama-7b](https://huggingface.co/codellama/CodeLlama-7b-hf) | [LLM]({{%relref "docs/features/text-generation" %}}) | ```docker run -ti -p 8080:8080 --gpus all localai/localai:{{< version >}}-cublas-cuda12 codellama-7b``` |
+| [codellama-7b-gguf](https://huggingface.co/TheBloke/CodeLlama-7B-GGUF) | [LLM]({{%relref "docs/features/text-generation" %}})  | ```docker run -ti -p 8080:8080 --gpus all localai/localai:{{< version >}}-cublas-cuda12-core codellama-7b-gguf``` |
 {{% /tab %}}
 
 {{< /tabs >}}
diff --git a/embedded/models/animagine-xl.yaml b/embedded/models/animagine-xl.yaml
new file mode 100644
index 00000000..d492c080
--- /dev/null
+++ b/embedded/models/animagine-xl.yaml
@@ -0,0 +1,17 @@
+name: animagine-xl
+parameters:
+  model: Linaqruf/animagine-xl
+backend: diffusers
+f16: true
+diffusers:
+  scheduler_type: euler_a
+
+usage: |
+        curl http://localhost:8080/v1/images/generations \
+          -H "Content-Type: application/json" \
+          -d '{
+            "prompt": "<positive prompt>|<negative prompt>",
+            "model": "animagine-xl",
+            "step": 51,
+            "size": "1024x1024"
+          }'
\ No newline at end of file
diff --git a/embedded/models/codellama-7b-gguf.yaml b/embedded/models/codellama-7b-gguf.yaml
new file mode 100644
index 00000000..d2a6b518
--- /dev/null
+++ b/embedded/models/codellama-7b-gguf.yaml
@@ -0,0 +1,16 @@
+name: codellama-7b-gguf
+backend: transformers
+parameters:
+  model: huggingface://TheBloke/CodeLlama-7B-GGUF/codellama-7b.Q4_K_M.gguf
+  temperature: 0.2
+  top_k: 40
+  seed: -1
+  top_p: 0.95
+context_size: 4096
+f16: true
+gpu_layers: 90
+usage: |
+      curl http://localhost:8080/v1/completions -H "Content-Type: application/json" -d '{
+          "model": "codellama-7b-gguf",
+          "prompt": "import socket\n\ndef ping_exponential_backoff(host: str):"
+      }'
\ No newline at end of file
diff --git a/embedded/models/codellama-7b.yaml b/embedded/models/codellama-7b.yaml
new file mode 100644
index 00000000..2dae7a8f
--- /dev/null
+++ b/embedded/models/codellama-7b.yaml
@@ -0,0 +1,14 @@
+name: codellama-7b
+backend: transformers
+parameters:
+  model: codellama/CodeLlama-7b-hf
+  temperature: 0.2
+  top_k: 40
+  seed: -1
+  top_p: 0.95
+
+usage: |
+      curl http://localhost:8080/v1/completions -H "Content-Type: application/json" -d '{
+          "model": "codellama-7b",
+          "prompt": "import socket\n\ndef ping_exponential_backoff(host: str):"
+      }'
\ No newline at end of file
diff --git a/embedded/models/dolphin-2.5-mixtral-8x7b.yaml b/embedded/models/dolphin-2.5-mixtral-8x7b.yaml
index b6df4799..ba6020c0 100644
--- a/embedded/models/dolphin-2.5-mixtral-8x7b.yaml
+++ b/embedded/models/dolphin-2.5-mixtral-8x7b.yaml
@@ -1,7 +1,7 @@
 name: dolphin-mixtral-8x7b
 mmap: true
 parameters:
-  model: huggingface://TheBloke/dolphin-2.5-mixtral-8x7b-GGUF/blob/main/dolphin-2.5-mixtral-8x7b.Q2_K.gguf
+  model: huggingface://TheBloke/dolphin-2.5-mixtral-8x7b-GGUF/dolphin-2.5-mixtral-8x7b.Q2_K.gguf
   temperature: 0.2
   top_k: 40
   top_p: 0.95
diff --git a/embedded/models/transformers-tinyllama.yaml b/embedded/models/transformers-tinyllama.yaml
new file mode 100644
index 00000000..0fcb64b6
--- /dev/null
+++ b/embedded/models/transformers-tinyllama.yaml
@@ -0,0 +1,32 @@
+name: tinyllama-chat
+backend: transformers
+type: AutoModelForCausalLM
+
+parameters:
+  model: TinyLlama/TinyLlama-1.1B-Chat-v1.0
+  temperature: 0.2
+  top_k: 40
+  seed: -1
+  top_p: 0.95
+  max_tokens: 4096
+
+template:
+  chat_message: |
+    <|im_start|>{{if eq .RoleName "assistant"}}assistant{{else if eq .RoleName "system"}}system{{else if eq .RoleName "user"}}user{{end}}
+    {{if .Content}}{{.Content}}{{end}}<|im_end|>
+  chat: |
+    {{.Input}}
+    <|im_start|>assistant
+    
+  completion: |
+    {{.Input}}
+
+stopwords:
+- <|im_end|>
+
+usage: |
+      curl http://localhost:8080/v1/chat/completions -H "Content-Type: application/json" -d '{
+        "model": "tinyllama-chat",
+        "messages": [{"role": "user", "content": "Say this is a test!"}],
+        "temperature": 0.7
+      }'
\ No newline at end of file
diff --git a/pkg/grpc/proto/backend.pb.go b/pkg/grpc/proto/backend.pb.go
index b9569785..acf9f375 100644
--- a/pkg/grpc/proto/backend.pb.go
+++ b/pkg/grpc/proto/backend.pb.go
@@ -1,7 +1,7 @@
 // Code generated by protoc-gen-go. DO NOT EDIT.
 // versions:
-// 	protoc-gen-go v1.28.1
-// 	protoc        v3.6.1
+// 	protoc-gen-go v1.26.0
+// 	protoc        v4.23.4
 // source: backend.proto
 
 package proto
@@ -567,8 +567,7 @@ type ModelOptions struct {
 	CLIPSubfolder string  `protobuf:"bytes,32,opt,name=CLIPSubfolder,proto3" json:"CLIPSubfolder,omitempty"`
 	CLIPSkip      int32   `protobuf:"varint,33,opt,name=CLIPSkip,proto3" json:"CLIPSkip,omitempty"`
 	ControlNet    string  `protobuf:"bytes,48,opt,name=ControlNet,proto3" json:"ControlNet,omitempty"`
-	// RWKV
-	Tokenizer string `protobuf:"bytes,34,opt,name=Tokenizer,proto3" json:"Tokenizer,omitempty"`
+	Tokenizer     string  `protobuf:"bytes,34,opt,name=Tokenizer,proto3" json:"Tokenizer,omitempty"`
 	// LLM (llama.cpp)
 	LoraBase    string  `protobuf:"bytes,35,opt,name=LoraBase,proto3" json:"LoraBase,omitempty"`
 	LoraAdapter string  `protobuf:"bytes,36,opt,name=LoraAdapter,proto3" json:"LoraAdapter,omitempty"`
@@ -584,6 +583,7 @@ type ModelOptions struct {
 	YarnAttnFactor float32 `protobuf:"fixed32,45,opt,name=YarnAttnFactor,proto3" json:"YarnAttnFactor,omitempty"`
 	YarnBetaFast   float32 `protobuf:"fixed32,46,opt,name=YarnBetaFast,proto3" json:"YarnBetaFast,omitempty"`
 	YarnBetaSlow   float32 `protobuf:"fixed32,47,opt,name=YarnBetaSlow,proto3" json:"YarnBetaSlow,omitempty"`
+	Type           string  `protobuf:"bytes,49,opt,name=Type,proto3" json:"Type,omitempty"`
 }
 
 func (x *ModelOptions) Reset() {
@@ -954,6 +954,13 @@ func (x *ModelOptions) GetYarnBetaSlow() float32 {
 	return 0
 }
 
+func (x *ModelOptions) GetType() string {
+	if x != nil {
+		return x.Type
+	}
+	return ""
+}
+
 type Result struct {
 	state         protoimpl.MessageState
 	sizeCache     protoimpl.SizeCache
@@ -1696,7 +1703,7 @@ var file_backend_proto_rawDesc = []byte{
 	0x65, 0x73, 0x18, 0x2a, 0x20, 0x03, 0x28, 0x09, 0x52, 0x06, 0x49, 0x6d, 0x61, 0x67, 0x65, 0x73,
 	0x22, 0x21, 0x0a, 0x05, 0x52, 0x65, 0x70, 0x6c, 0x79, 0x12, 0x18, 0x0a, 0x07, 0x6d, 0x65, 0x73,
 	0x73, 0x61, 0x67, 0x65, 0x18, 0x01, 0x20, 0x01, 0x28, 0x0c, 0x52, 0x07, 0x6d, 0x65, 0x73, 0x73,
-	0x61, 0x67, 0x65, 0x22, 0xcc, 0x0b, 0x0a, 0x0c, 0x4d, 0x6f, 0x64, 0x65, 0x6c, 0x4f, 0x70, 0x74,
+	0x61, 0x67, 0x65, 0x22, 0xe0, 0x0b, 0x0a, 0x0c, 0x4d, 0x6f, 0x64, 0x65, 0x6c, 0x4f, 0x70, 0x74,
 	0x69, 0x6f, 0x6e, 0x73, 0x12, 0x14, 0x0a, 0x05, 0x4d, 0x6f, 0x64, 0x65, 0x6c, 0x18, 0x01, 0x20,
 	0x01, 0x28, 0x09, 0x52, 0x05, 0x4d, 0x6f, 0x64, 0x65, 0x6c, 0x12, 0x20, 0x0a, 0x0b, 0x43, 0x6f,
 	0x6e, 0x74, 0x65, 0x78, 0x74, 0x53, 0x69, 0x7a, 0x65, 0x18, 0x02, 0x20, 0x01, 0x28, 0x05, 0x52,
@@ -1789,131 +1796,132 @@ var file_backend_proto_rawDesc = []byte{
 	0x52, 0x0c, 0x59, 0x61, 0x72, 0x6e, 0x42, 0x65, 0x74, 0x61, 0x46, 0x61, 0x73, 0x74, 0x12, 0x22,
 	0x0a, 0x0c, 0x59, 0x61, 0x72, 0x6e, 0x42, 0x65, 0x74, 0x61, 0x53, 0x6c, 0x6f, 0x77, 0x18, 0x2f,
 	0x20, 0x01, 0x28, 0x02, 0x52, 0x0c, 0x59, 0x61, 0x72, 0x6e, 0x42, 0x65, 0x74, 0x61, 0x53, 0x6c,
-	0x6f, 0x77, 0x22, 0x3c, 0x0a, 0x06, 0x52, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x12, 0x18, 0x0a, 0x07,
-	0x6d, 0x65, 0x73, 0x73, 0x61, 0x67, 0x65, 0x18, 0x01, 0x20, 0x01, 0x28, 0x09, 0x52, 0x07, 0x6d,
-	0x65, 0x73, 0x73, 0x61, 0x67, 0x65, 0x12, 0x18, 0x0a, 0x07, 0x73, 0x75, 0x63, 0x63, 0x65, 0x73,
-	0x73, 0x18, 0x02, 0x20, 0x01, 0x28, 0x08, 0x52, 0x07, 0x73, 0x75, 0x63, 0x63, 0x65, 0x73, 0x73,
-	0x22, 0x31, 0x0a, 0x0f, 0x45, 0x6d, 0x62, 0x65, 0x64, 0x64, 0x69, 0x6e, 0x67, 0x52, 0x65, 0x73,
-	0x75, 0x6c, 0x74, 0x12, 0x1e, 0x0a, 0x0a, 0x65, 0x6d, 0x62, 0x65, 0x64, 0x64, 0x69, 0x6e, 0x67,
-	0x73, 0x18, 0x01, 0x20, 0x03, 0x28, 0x02, 0x52, 0x0a, 0x65, 0x6d, 0x62, 0x65, 0x64, 0x64, 0x69,
-	0x6e, 0x67, 0x73, 0x22, 0x5b, 0x0a, 0x11, 0x54, 0x72, 0x61, 0x6e, 0x73, 0x63, 0x72, 0x69, 0x70,
-	0x74, 0x52, 0x65, 0x71, 0x75, 0x65, 0x73, 0x74, 0x12, 0x10, 0x0a, 0x03, 0x64, 0x73, 0x74, 0x18,
-	0x02, 0x20, 0x01, 0x28, 0x09, 0x52, 0x03, 0x64, 0x73, 0x74, 0x12, 0x1a, 0x0a, 0x08, 0x6c, 0x61,
-	0x6e, 0x67, 0x75, 0x61, 0x67, 0x65, 0x18, 0x03, 0x20, 0x01, 0x28, 0x09, 0x52, 0x08, 0x6c, 0x61,
-	0x6e, 0x67, 0x75, 0x61, 0x67, 0x65, 0x12, 0x18, 0x0a, 0x07, 0x74, 0x68, 0x72, 0x65, 0x61, 0x64,
-	0x73, 0x18, 0x04, 0x20, 0x01, 0x28, 0x0d, 0x52, 0x07, 0x74, 0x68, 0x72, 0x65, 0x61, 0x64, 0x73,
-	0x22, 0x5e, 0x0a, 0x10, 0x54, 0x72, 0x61, 0x6e, 0x73, 0x63, 0x72, 0x69, 0x70, 0x74, 0x52, 0x65,
-	0x73, 0x75, 0x6c, 0x74, 0x12, 0x36, 0x0a, 0x08, 0x73, 0x65, 0x67, 0x6d, 0x65, 0x6e, 0x74, 0x73,
-	0x18, 0x01, 0x20, 0x03, 0x28, 0x0b, 0x32, 0x1a, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64,
-	0x2e, 0x54, 0x72, 0x61, 0x6e, 0x73, 0x63, 0x72, 0x69, 0x70, 0x74, 0x53, 0x65, 0x67, 0x6d, 0x65,
-	0x6e, 0x74, 0x52, 0x08, 0x73, 0x65, 0x67, 0x6d, 0x65, 0x6e, 0x74, 0x73, 0x12, 0x12, 0x0a, 0x04,
-	0x74, 0x65, 0x78, 0x74, 0x18, 0x02, 0x20, 0x01, 0x28, 0x09, 0x52, 0x04, 0x74, 0x65, 0x78, 0x74,
-	0x22, 0x77, 0x0a, 0x11, 0x54, 0x72, 0x61, 0x6e, 0x73, 0x63, 0x72, 0x69, 0x70, 0x74, 0x53, 0x65,
-	0x67, 0x6d, 0x65, 0x6e, 0x74, 0x12, 0x0e, 0x0a, 0x02, 0x69, 0x64, 0x18, 0x01, 0x20, 0x01, 0x28,
-	0x05, 0x52, 0x02, 0x69, 0x64, 0x12, 0x14, 0x0a, 0x05, 0x73, 0x74, 0x61, 0x72, 0x74, 0x18, 0x02,
-	0x20, 0x01, 0x28, 0x03, 0x52, 0x05, 0x73, 0x74, 0x61, 0x72, 0x74, 0x12, 0x10, 0x0a, 0x03, 0x65,
-	0x6e, 0x64, 0x18, 0x03, 0x20, 0x01, 0x28, 0x03, 0x52, 0x03, 0x65, 0x6e, 0x64, 0x12, 0x12, 0x0a,
-	0x04, 0x74, 0x65, 0x78, 0x74, 0x18, 0x04, 0x20, 0x01, 0x28, 0x09, 0x52, 0x04, 0x74, 0x65, 0x78,
-	0x74, 0x12, 0x16, 0x0a, 0x06, 0x74, 0x6f, 0x6b, 0x65, 0x6e, 0x73, 0x18, 0x05, 0x20, 0x03, 0x28,
-	0x05, 0x52, 0x06, 0x74, 0x6f, 0x6b, 0x65, 0x6e, 0x73, 0x22, 0xbe, 0x02, 0x0a, 0x14, 0x47, 0x65,
-	0x6e, 0x65, 0x72, 0x61, 0x74, 0x65, 0x49, 0x6d, 0x61, 0x67, 0x65, 0x52, 0x65, 0x71, 0x75, 0x65,
-	0x73, 0x74, 0x12, 0x16, 0x0a, 0x06, 0x68, 0x65, 0x69, 0x67, 0x68, 0x74, 0x18, 0x01, 0x20, 0x01,
-	0x28, 0x05, 0x52, 0x06, 0x68, 0x65, 0x69, 0x67, 0x68, 0x74, 0x12, 0x14, 0x0a, 0x05, 0x77, 0x69,
-	0x64, 0x74, 0x68, 0x18, 0x02, 0x20, 0x01, 0x28, 0x05, 0x52, 0x05, 0x77, 0x69, 0x64, 0x74, 0x68,
-	0x12, 0x12, 0x0a, 0x04, 0x6d, 0x6f, 0x64, 0x65, 0x18, 0x03, 0x20, 0x01, 0x28, 0x05, 0x52, 0x04,
-	0x6d, 0x6f, 0x64, 0x65, 0x12, 0x12, 0x0a, 0x04, 0x73, 0x74, 0x65, 0x70, 0x18, 0x04, 0x20, 0x01,
-	0x28, 0x05, 0x52, 0x04, 0x73, 0x74, 0x65, 0x70, 0x12, 0x12, 0x0a, 0x04, 0x73, 0x65, 0x65, 0x64,
-	0x18, 0x05, 0x20, 0x01, 0x28, 0x05, 0x52, 0x04, 0x73, 0x65, 0x65, 0x64, 0x12, 0x27, 0x0a, 0x0f,
-	0x70, 0x6f, 0x73, 0x69, 0x74, 0x69, 0x76, 0x65, 0x5f, 0x70, 0x72, 0x6f, 0x6d, 0x70, 0x74, 0x18,
-	0x06, 0x20, 0x01, 0x28, 0x09, 0x52, 0x0e, 0x70, 0x6f, 0x73, 0x69, 0x74, 0x69, 0x76, 0x65, 0x50,
-	0x72, 0x6f, 0x6d, 0x70, 0x74, 0x12, 0x27, 0x0a, 0x0f, 0x6e, 0x65, 0x67, 0x61, 0x74, 0x69, 0x76,
-	0x65, 0x5f, 0x70, 0x72, 0x6f, 0x6d, 0x70, 0x74, 0x18, 0x07, 0x20, 0x01, 0x28, 0x09, 0x52, 0x0e,
-	0x6e, 0x65, 0x67, 0x61, 0x74, 0x69, 0x76, 0x65, 0x50, 0x72, 0x6f, 0x6d, 0x70, 0x74, 0x12, 0x10,
-	0x0a, 0x03, 0x64, 0x73, 0x74, 0x18, 0x08, 0x20, 0x01, 0x28, 0x09, 0x52, 0x03, 0x64, 0x73, 0x74,
-	0x12, 0x10, 0x0a, 0x03, 0x73, 0x72, 0x63, 0x18, 0x09, 0x20, 0x01, 0x28, 0x09, 0x52, 0x03, 0x73,
-	0x72, 0x63, 0x12, 0x2a, 0x0a, 0x10, 0x45, 0x6e, 0x61, 0x62, 0x6c, 0x65, 0x50, 0x61, 0x72, 0x61,
-	0x6d, 0x65, 0x74, 0x65, 0x72, 0x73, 0x18, 0x0a, 0x20, 0x01, 0x28, 0x09, 0x52, 0x10, 0x45, 0x6e,
-	0x61, 0x62, 0x6c, 0x65, 0x50, 0x61, 0x72, 0x61, 0x6d, 0x65, 0x74, 0x65, 0x72, 0x73, 0x12, 0x1a,
-	0x0a, 0x08, 0x43, 0x4c, 0x49, 0x50, 0x53, 0x6b, 0x69, 0x70, 0x18, 0x0b, 0x20, 0x01, 0x28, 0x05,
-	0x52, 0x08, 0x43, 0x4c, 0x49, 0x50, 0x53, 0x6b, 0x69, 0x70, 0x22, 0x48, 0x0a, 0x0a, 0x54, 0x54,
-	0x53, 0x52, 0x65, 0x71, 0x75, 0x65, 0x73, 0x74, 0x12, 0x12, 0x0a, 0x04, 0x74, 0x65, 0x78, 0x74,
-	0x18, 0x01, 0x20, 0x01, 0x28, 0x09, 0x52, 0x04, 0x74, 0x65, 0x78, 0x74, 0x12, 0x14, 0x0a, 0x05,
-	0x6d, 0x6f, 0x64, 0x65, 0x6c, 0x18, 0x02, 0x20, 0x01, 0x28, 0x09, 0x52, 0x05, 0x6d, 0x6f, 0x64,
-	0x65, 0x6c, 0x12, 0x10, 0x0a, 0x03, 0x64, 0x73, 0x74, 0x18, 0x03, 0x20, 0x01, 0x28, 0x09, 0x52,
-	0x03, 0x64, 0x73, 0x74, 0x22, 0x46, 0x0a, 0x14, 0x54, 0x6f, 0x6b, 0x65, 0x6e, 0x69, 0x7a, 0x61,
-	0x74, 0x69, 0x6f, 0x6e, 0x52, 0x65, 0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65, 0x12, 0x16, 0x0a, 0x06,
-	0x6c, 0x65, 0x6e, 0x67, 0x74, 0x68, 0x18, 0x01, 0x20, 0x01, 0x28, 0x05, 0x52, 0x06, 0x6c, 0x65,
-	0x6e, 0x67, 0x74, 0x68, 0x12, 0x16, 0x0a, 0x06, 0x74, 0x6f, 0x6b, 0x65, 0x6e, 0x73, 0x18, 0x02,
-	0x20, 0x03, 0x28, 0x05, 0x52, 0x06, 0x74, 0x6f, 0x6b, 0x65, 0x6e, 0x73, 0x22, 0xac, 0x01, 0x0a,
-	0x0f, 0x4d, 0x65, 0x6d, 0x6f, 0x72, 0x79, 0x55, 0x73, 0x61, 0x67, 0x65, 0x44, 0x61, 0x74, 0x61,
-	0x12, 0x14, 0x0a, 0x05, 0x74, 0x6f, 0x74, 0x61, 0x6c, 0x18, 0x01, 0x20, 0x01, 0x28, 0x04, 0x52,
-	0x05, 0x74, 0x6f, 0x74, 0x61, 0x6c, 0x12, 0x45, 0x0a, 0x09, 0x62, 0x72, 0x65, 0x61, 0x6b, 0x64,
-	0x6f, 0x77, 0x6e, 0x18, 0x02, 0x20, 0x03, 0x28, 0x0b, 0x32, 0x27, 0x2e, 0x62, 0x61, 0x63, 0x6b,
-	0x65, 0x6e, 0x64, 0x2e, 0x4d, 0x65, 0x6d, 0x6f, 0x72, 0x79, 0x55, 0x73, 0x61, 0x67, 0x65, 0x44,
-	0x61, 0x74, 0x61, 0x2e, 0x42, 0x72, 0x65, 0x61, 0x6b, 0x64, 0x6f, 0x77, 0x6e, 0x45, 0x6e, 0x74,
-	0x72, 0x79, 0x52, 0x09, 0x62, 0x72, 0x65, 0x61, 0x6b, 0x64, 0x6f, 0x77, 0x6e, 0x1a, 0x3c, 0x0a,
-	0x0e, 0x42, 0x72, 0x65, 0x61, 0x6b, 0x64, 0x6f, 0x77, 0x6e, 0x45, 0x6e, 0x74, 0x72, 0x79, 0x12,
-	0x10, 0x0a, 0x03, 0x6b, 0x65, 0x79, 0x18, 0x01, 0x20, 0x01, 0x28, 0x09, 0x52, 0x03, 0x6b, 0x65,
-	0x79, 0x12, 0x14, 0x0a, 0x05, 0x76, 0x61, 0x6c, 0x75, 0x65, 0x18, 0x02, 0x20, 0x01, 0x28, 0x04,
-	0x52, 0x05, 0x76, 0x61, 0x6c, 0x75, 0x65, 0x3a, 0x02, 0x38, 0x01, 0x22, 0xbc, 0x01, 0x0a, 0x0e,
-	0x53, 0x74, 0x61, 0x74, 0x75, 0x73, 0x52, 0x65, 0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65, 0x12, 0x33,
-	0x0a, 0x05, 0x73, 0x74, 0x61, 0x74, 0x65, 0x18, 0x01, 0x20, 0x01, 0x28, 0x0e, 0x32, 0x1d, 0x2e,
-	0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x53, 0x74, 0x61, 0x74, 0x75, 0x73, 0x52, 0x65,
-	0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65, 0x2e, 0x53, 0x74, 0x61, 0x74, 0x65, 0x52, 0x05, 0x73, 0x74,
-	0x61, 0x74, 0x65, 0x12, 0x30, 0x0a, 0x06, 0x6d, 0x65, 0x6d, 0x6f, 0x72, 0x79, 0x18, 0x02, 0x20,
-	0x01, 0x28, 0x0b, 0x32, 0x18, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x4d, 0x65,
-	0x6d, 0x6f, 0x72, 0x79, 0x55, 0x73, 0x61, 0x67, 0x65, 0x44, 0x61, 0x74, 0x61, 0x52, 0x06, 0x6d,
-	0x65, 0x6d, 0x6f, 0x72, 0x79, 0x22, 0x43, 0x0a, 0x05, 0x53, 0x74, 0x61, 0x74, 0x65, 0x12, 0x11,
-	0x0a, 0x0d, 0x55, 0x4e, 0x49, 0x4e, 0x49, 0x54, 0x49, 0x41, 0x4c, 0x49, 0x5a, 0x45, 0x44, 0x10,
-	0x00, 0x12, 0x08, 0x0a, 0x04, 0x42, 0x55, 0x53, 0x59, 0x10, 0x01, 0x12, 0x09, 0x0a, 0x05, 0x52,
-	0x45, 0x41, 0x44, 0x59, 0x10, 0x02, 0x12, 0x12, 0x0a, 0x05, 0x45, 0x52, 0x52, 0x4f, 0x52, 0x10,
-	0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x01, 0x32, 0xf4, 0x04, 0x0a, 0x07, 0x42,
-	0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x12, 0x32, 0x0a, 0x06, 0x48, 0x65, 0x61, 0x6c, 0x74, 0x68,
-	0x12, 0x16, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x48, 0x65, 0x61, 0x6c, 0x74,
-	0x68, 0x4d, 0x65, 0x73, 0x73, 0x61, 0x67, 0x65, 0x1a, 0x0e, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65,
-	0x6e, 0x64, 0x2e, 0x52, 0x65, 0x70, 0x6c, 0x79, 0x22, 0x00, 0x12, 0x34, 0x0a, 0x07, 0x50, 0x72,
-	0x65, 0x64, 0x69, 0x63, 0x74, 0x12, 0x17, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e,
-	0x50, 0x72, 0x65, 0x64, 0x69, 0x63, 0x74, 0x4f, 0x70, 0x74, 0x69, 0x6f, 0x6e, 0x73, 0x1a, 0x0e,
-	0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x52, 0x65, 0x70, 0x6c, 0x79, 0x22, 0x00,
-	0x12, 0x35, 0x0a, 0x09, 0x4c, 0x6f, 0x61, 0x64, 0x4d, 0x6f, 0x64, 0x65, 0x6c, 0x12, 0x15, 0x2e,
-	0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x4d, 0x6f, 0x64, 0x65, 0x6c, 0x4f, 0x70, 0x74,
-	0x69, 0x6f, 0x6e, 0x73, 0x1a, 0x0f, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x52,
-	0x65, 0x73, 0x75, 0x6c, 0x74, 0x22, 0x00, 0x12, 0x3c, 0x0a, 0x0d, 0x50, 0x72, 0x65, 0x64, 0x69,
-	0x63, 0x74, 0x53, 0x74, 0x72, 0x65, 0x61, 0x6d, 0x12, 0x17, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65,
-	0x6e, 0x64, 0x2e, 0x50, 0x72, 0x65, 0x64, 0x69, 0x63, 0x74, 0x4f, 0x70, 0x74, 0x69, 0x6f, 0x6e,
-	0x73, 0x1a, 0x0e, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x52, 0x65, 0x70, 0x6c,
-	0x79, 0x22, 0x00, 0x30, 0x01, 0x12, 0x40, 0x0a, 0x09, 0x45, 0x6d, 0x62, 0x65, 0x64, 0x64, 0x69,
-	0x6e, 0x67, 0x12, 0x17, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x50, 0x72, 0x65,
-	0x64, 0x69, 0x63, 0x74, 0x4f, 0x70, 0x74, 0x69, 0x6f, 0x6e, 0x73, 0x1a, 0x18, 0x2e, 0x62, 0x61,
-	0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x45, 0x6d, 0x62, 0x65, 0x64, 0x64, 0x69, 0x6e, 0x67, 0x52,
-	0x65, 0x73, 0x75, 0x6c, 0x74, 0x22, 0x00, 0x12, 0x41, 0x0a, 0x0d, 0x47, 0x65, 0x6e, 0x65, 0x72,
-	0x61, 0x74, 0x65, 0x49, 0x6d, 0x61, 0x67, 0x65, 0x12, 0x1d, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65,
-	0x6e, 0x64, 0x2e, 0x47, 0x65, 0x6e, 0x65, 0x72, 0x61, 0x74, 0x65, 0x49, 0x6d, 0x61, 0x67, 0x65,
-	0x52, 0x65, 0x71, 0x75, 0x65, 0x73, 0x74, 0x1a, 0x0f, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e,
-	0x64, 0x2e, 0x52, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x22, 0x00, 0x12, 0x4d, 0x0a, 0x12, 0x41, 0x75,
-	0x64, 0x69, 0x6f, 0x54, 0x72, 0x61, 0x6e, 0x73, 0x63, 0x72, 0x69, 0x70, 0x74, 0x69, 0x6f, 0x6e,
-	0x12, 0x1a, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x54, 0x72, 0x61, 0x6e, 0x73,
-	0x63, 0x72, 0x69, 0x70, 0x74, 0x52, 0x65, 0x71, 0x75, 0x65, 0x73, 0x74, 0x1a, 0x19, 0x2e, 0x62,
-	0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x54, 0x72, 0x61, 0x6e, 0x73, 0x63, 0x72, 0x69, 0x70,
-	0x74, 0x52, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x22, 0x00, 0x12, 0x2d, 0x0a, 0x03, 0x54, 0x54, 0x53,
-	0x12, 0x13, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x54, 0x54, 0x53, 0x52, 0x65,
-	0x71, 0x75, 0x65, 0x73, 0x74, 0x1a, 0x0f, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e,
-	0x52, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x22, 0x00, 0x12, 0x4a, 0x0a, 0x0e, 0x54, 0x6f, 0x6b, 0x65,
-	0x6e, 0x69, 0x7a, 0x65, 0x53, 0x74, 0x72, 0x69, 0x6e, 0x67, 0x12, 0x17, 0x2e, 0x62, 0x61, 0x63,
-	0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x50, 0x72, 0x65, 0x64, 0x69, 0x63, 0x74, 0x4f, 0x70, 0x74, 0x69,
-	0x6f, 0x6e, 0x73, 0x1a, 0x1d, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x54, 0x6f,
-	0x6b, 0x65, 0x6e, 0x69, 0x7a, 0x61, 0x74, 0x69, 0x6f, 0x6e, 0x52, 0x65, 0x73, 0x70, 0x6f, 0x6e,
-	0x73, 0x65, 0x22, 0x00, 0x12, 0x3b, 0x0a, 0x06, 0x53, 0x74, 0x61, 0x74, 0x75, 0x73, 0x12, 0x16,
-	0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x48, 0x65, 0x61, 0x6c, 0x74, 0x68, 0x4d,
-	0x65, 0x73, 0x73, 0x61, 0x67, 0x65, 0x1a, 0x17, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64,
-	0x2e, 0x53, 0x74, 0x61, 0x74, 0x75, 0x73, 0x52, 0x65, 0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65, 0x22,
-	0x00, 0x42, 0x5a, 0x0a, 0x19, 0x69, 0x6f, 0x2e, 0x73, 0x6b, 0x79, 0x6e, 0x65, 0x74, 0x2e, 0x6c,
-	0x6f, 0x63, 0x61, 0x6c, 0x61, 0x69, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x42, 0x0e,
-	0x4c, 0x6f, 0x63, 0x61, 0x6c, 0x41, 0x49, 0x42, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x50, 0x01,
-	0x5a, 0x2b, 0x67, 0x69, 0x74, 0x68, 0x75, 0x62, 0x2e, 0x63, 0x6f, 0x6d, 0x2f, 0x67, 0x6f, 0x2d,
-	0x73, 0x6b, 0x79, 0x6e, 0x65, 0x74, 0x2f, 0x4c, 0x6f, 0x63, 0x61, 0x6c, 0x41, 0x49, 0x2f, 0x70,
-	0x6b, 0x67, 0x2f, 0x67, 0x72, 0x70, 0x63, 0x2f, 0x70, 0x72, 0x6f, 0x74, 0x6f, 0x62, 0x06, 0x70,
-	0x72, 0x6f, 0x74, 0x6f, 0x33,
+	0x6f, 0x77, 0x12, 0x12, 0x0a, 0x04, 0x54, 0x79, 0x70, 0x65, 0x18, 0x31, 0x20, 0x01, 0x28, 0x09,
+	0x52, 0x04, 0x54, 0x79, 0x70, 0x65, 0x22, 0x3c, 0x0a, 0x06, 0x52, 0x65, 0x73, 0x75, 0x6c, 0x74,
+	0x12, 0x18, 0x0a, 0x07, 0x6d, 0x65, 0x73, 0x73, 0x61, 0x67, 0x65, 0x18, 0x01, 0x20, 0x01, 0x28,
+	0x09, 0x52, 0x07, 0x6d, 0x65, 0x73, 0x73, 0x61, 0x67, 0x65, 0x12, 0x18, 0x0a, 0x07, 0x73, 0x75,
+	0x63, 0x63, 0x65, 0x73, 0x73, 0x18, 0x02, 0x20, 0x01, 0x28, 0x08, 0x52, 0x07, 0x73, 0x75, 0x63,
+	0x63, 0x65, 0x73, 0x73, 0x22, 0x31, 0x0a, 0x0f, 0x45, 0x6d, 0x62, 0x65, 0x64, 0x64, 0x69, 0x6e,
+	0x67, 0x52, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x12, 0x1e, 0x0a, 0x0a, 0x65, 0x6d, 0x62, 0x65, 0x64,
+	0x64, 0x69, 0x6e, 0x67, 0x73, 0x18, 0x01, 0x20, 0x03, 0x28, 0x02, 0x52, 0x0a, 0x65, 0x6d, 0x62,
+	0x65, 0x64, 0x64, 0x69, 0x6e, 0x67, 0x73, 0x22, 0x5b, 0x0a, 0x11, 0x54, 0x72, 0x61, 0x6e, 0x73,
+	0x63, 0x72, 0x69, 0x70, 0x74, 0x52, 0x65, 0x71, 0x75, 0x65, 0x73, 0x74, 0x12, 0x10, 0x0a, 0x03,
+	0x64, 0x73, 0x74, 0x18, 0x02, 0x20, 0x01, 0x28, 0x09, 0x52, 0x03, 0x64, 0x73, 0x74, 0x12, 0x1a,
+	0x0a, 0x08, 0x6c, 0x61, 0x6e, 0x67, 0x75, 0x61, 0x67, 0x65, 0x18, 0x03, 0x20, 0x01, 0x28, 0x09,
+	0x52, 0x08, 0x6c, 0x61, 0x6e, 0x67, 0x75, 0x61, 0x67, 0x65, 0x12, 0x18, 0x0a, 0x07, 0x74, 0x68,
+	0x72, 0x65, 0x61, 0x64, 0x73, 0x18, 0x04, 0x20, 0x01, 0x28, 0x0d, 0x52, 0x07, 0x74, 0x68, 0x72,
+	0x65, 0x61, 0x64, 0x73, 0x22, 0x5e, 0x0a, 0x10, 0x54, 0x72, 0x61, 0x6e, 0x73, 0x63, 0x72, 0x69,
+	0x70, 0x74, 0x52, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x12, 0x36, 0x0a, 0x08, 0x73, 0x65, 0x67, 0x6d,
+	0x65, 0x6e, 0x74, 0x73, 0x18, 0x01, 0x20, 0x03, 0x28, 0x0b, 0x32, 0x1a, 0x2e, 0x62, 0x61, 0x63,
+	0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x54, 0x72, 0x61, 0x6e, 0x73, 0x63, 0x72, 0x69, 0x70, 0x74, 0x53,
+	0x65, 0x67, 0x6d, 0x65, 0x6e, 0x74, 0x52, 0x08, 0x73, 0x65, 0x67, 0x6d, 0x65, 0x6e, 0x74, 0x73,
+	0x12, 0x12, 0x0a, 0x04, 0x74, 0x65, 0x78, 0x74, 0x18, 0x02, 0x20, 0x01, 0x28, 0x09, 0x52, 0x04,
+	0x74, 0x65, 0x78, 0x74, 0x22, 0x77, 0x0a, 0x11, 0x54, 0x72, 0x61, 0x6e, 0x73, 0x63, 0x72, 0x69,
+	0x70, 0x74, 0x53, 0x65, 0x67, 0x6d, 0x65, 0x6e, 0x74, 0x12, 0x0e, 0x0a, 0x02, 0x69, 0x64, 0x18,
+	0x01, 0x20, 0x01, 0x28, 0x05, 0x52, 0x02, 0x69, 0x64, 0x12, 0x14, 0x0a, 0x05, 0x73, 0x74, 0x61,
+	0x72, 0x74, 0x18, 0x02, 0x20, 0x01, 0x28, 0x03, 0x52, 0x05, 0x73, 0x74, 0x61, 0x72, 0x74, 0x12,
+	0x10, 0x0a, 0x03, 0x65, 0x6e, 0x64, 0x18, 0x03, 0x20, 0x01, 0x28, 0x03, 0x52, 0x03, 0x65, 0x6e,
+	0x64, 0x12, 0x12, 0x0a, 0x04, 0x74, 0x65, 0x78, 0x74, 0x18, 0x04, 0x20, 0x01, 0x28, 0x09, 0x52,
+	0x04, 0x74, 0x65, 0x78, 0x74, 0x12, 0x16, 0x0a, 0x06, 0x74, 0x6f, 0x6b, 0x65, 0x6e, 0x73, 0x18,
+	0x05, 0x20, 0x03, 0x28, 0x05, 0x52, 0x06, 0x74, 0x6f, 0x6b, 0x65, 0x6e, 0x73, 0x22, 0xbe, 0x02,
+	0x0a, 0x14, 0x47, 0x65, 0x6e, 0x65, 0x72, 0x61, 0x74, 0x65, 0x49, 0x6d, 0x61, 0x67, 0x65, 0x52,
+	0x65, 0x71, 0x75, 0x65, 0x73, 0x74, 0x12, 0x16, 0x0a, 0x06, 0x68, 0x65, 0x69, 0x67, 0x68, 0x74,
+	0x18, 0x01, 0x20, 0x01, 0x28, 0x05, 0x52, 0x06, 0x68, 0x65, 0x69, 0x67, 0x68, 0x74, 0x12, 0x14,
+	0x0a, 0x05, 0x77, 0x69, 0x64, 0x74, 0x68, 0x18, 0x02, 0x20, 0x01, 0x28, 0x05, 0x52, 0x05, 0x77,
+	0x69, 0x64, 0x74, 0x68, 0x12, 0x12, 0x0a, 0x04, 0x6d, 0x6f, 0x64, 0x65, 0x18, 0x03, 0x20, 0x01,
+	0x28, 0x05, 0x52, 0x04, 0x6d, 0x6f, 0x64, 0x65, 0x12, 0x12, 0x0a, 0x04, 0x73, 0x74, 0x65, 0x70,
+	0x18, 0x04, 0x20, 0x01, 0x28, 0x05, 0x52, 0x04, 0x73, 0x74, 0x65, 0x70, 0x12, 0x12, 0x0a, 0x04,
+	0x73, 0x65, 0x65, 0x64, 0x18, 0x05, 0x20, 0x01, 0x28, 0x05, 0x52, 0x04, 0x73, 0x65, 0x65, 0x64,
+	0x12, 0x27, 0x0a, 0x0f, 0x70, 0x6f, 0x73, 0x69, 0x74, 0x69, 0x76, 0x65, 0x5f, 0x70, 0x72, 0x6f,
+	0x6d, 0x70, 0x74, 0x18, 0x06, 0x20, 0x01, 0x28, 0x09, 0x52, 0x0e, 0x70, 0x6f, 0x73, 0x69, 0x74,
+	0x69, 0x76, 0x65, 0x50, 0x72, 0x6f, 0x6d, 0x70, 0x74, 0x12, 0x27, 0x0a, 0x0f, 0x6e, 0x65, 0x67,
+	0x61, 0x74, 0x69, 0x76, 0x65, 0x5f, 0x70, 0x72, 0x6f, 0x6d, 0x70, 0x74, 0x18, 0x07, 0x20, 0x01,
+	0x28, 0x09, 0x52, 0x0e, 0x6e, 0x65, 0x67, 0x61, 0x74, 0x69, 0x76, 0x65, 0x50, 0x72, 0x6f, 0x6d,
+	0x70, 0x74, 0x12, 0x10, 0x0a, 0x03, 0x64, 0x73, 0x74, 0x18, 0x08, 0x20, 0x01, 0x28, 0x09, 0x52,
+	0x03, 0x64, 0x73, 0x74, 0x12, 0x10, 0x0a, 0x03, 0x73, 0x72, 0x63, 0x18, 0x09, 0x20, 0x01, 0x28,
+	0x09, 0x52, 0x03, 0x73, 0x72, 0x63, 0x12, 0x2a, 0x0a, 0x10, 0x45, 0x6e, 0x61, 0x62, 0x6c, 0x65,
+	0x50, 0x61, 0x72, 0x61, 0x6d, 0x65, 0x74, 0x65, 0x72, 0x73, 0x18, 0x0a, 0x20, 0x01, 0x28, 0x09,
+	0x52, 0x10, 0x45, 0x6e, 0x61, 0x62, 0x6c, 0x65, 0x50, 0x61, 0x72, 0x61, 0x6d, 0x65, 0x74, 0x65,
+	0x72, 0x73, 0x12, 0x1a, 0x0a, 0x08, 0x43, 0x4c, 0x49, 0x50, 0x53, 0x6b, 0x69, 0x70, 0x18, 0x0b,
+	0x20, 0x01, 0x28, 0x05, 0x52, 0x08, 0x43, 0x4c, 0x49, 0x50, 0x53, 0x6b, 0x69, 0x70, 0x22, 0x48,
+	0x0a, 0x0a, 0x54, 0x54, 0x53, 0x52, 0x65, 0x71, 0x75, 0x65, 0x73, 0x74, 0x12, 0x12, 0x0a, 0x04,
+	0x74, 0x65, 0x78, 0x74, 0x18, 0x01, 0x20, 0x01, 0x28, 0x09, 0x52, 0x04, 0x74, 0x65, 0x78, 0x74,
+	0x12, 0x14, 0x0a, 0x05, 0x6d, 0x6f, 0x64, 0x65, 0x6c, 0x18, 0x02, 0x20, 0x01, 0x28, 0x09, 0x52,
+	0x05, 0x6d, 0x6f, 0x64, 0x65, 0x6c, 0x12, 0x10, 0x0a, 0x03, 0x64, 0x73, 0x74, 0x18, 0x03, 0x20,
+	0x01, 0x28, 0x09, 0x52, 0x03, 0x64, 0x73, 0x74, 0x22, 0x46, 0x0a, 0x14, 0x54, 0x6f, 0x6b, 0x65,
+	0x6e, 0x69, 0x7a, 0x61, 0x74, 0x69, 0x6f, 0x6e, 0x52, 0x65, 0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65,
+	0x12, 0x16, 0x0a, 0x06, 0x6c, 0x65, 0x6e, 0x67, 0x74, 0x68, 0x18, 0x01, 0x20, 0x01, 0x28, 0x05,
+	0x52, 0x06, 0x6c, 0x65, 0x6e, 0x67, 0x74, 0x68, 0x12, 0x16, 0x0a, 0x06, 0x74, 0x6f, 0x6b, 0x65,
+	0x6e, 0x73, 0x18, 0x02, 0x20, 0x03, 0x28, 0x05, 0x52, 0x06, 0x74, 0x6f, 0x6b, 0x65, 0x6e, 0x73,
+	0x22, 0xac, 0x01, 0x0a, 0x0f, 0x4d, 0x65, 0x6d, 0x6f, 0x72, 0x79, 0x55, 0x73, 0x61, 0x67, 0x65,
+	0x44, 0x61, 0x74, 0x61, 0x12, 0x14, 0x0a, 0x05, 0x74, 0x6f, 0x74, 0x61, 0x6c, 0x18, 0x01, 0x20,
+	0x01, 0x28, 0x04, 0x52, 0x05, 0x74, 0x6f, 0x74, 0x61, 0x6c, 0x12, 0x45, 0x0a, 0x09, 0x62, 0x72,
+	0x65, 0x61, 0x6b, 0x64, 0x6f, 0x77, 0x6e, 0x18, 0x02, 0x20, 0x03, 0x28, 0x0b, 0x32, 0x27, 0x2e,
+	0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x4d, 0x65, 0x6d, 0x6f, 0x72, 0x79, 0x55, 0x73,
+	0x61, 0x67, 0x65, 0x44, 0x61, 0x74, 0x61, 0x2e, 0x42, 0x72, 0x65, 0x61, 0x6b, 0x64, 0x6f, 0x77,
+	0x6e, 0x45, 0x6e, 0x74, 0x72, 0x79, 0x52, 0x09, 0x62, 0x72, 0x65, 0x61, 0x6b, 0x64, 0x6f, 0x77,
+	0x6e, 0x1a, 0x3c, 0x0a, 0x0e, 0x42, 0x72, 0x65, 0x61, 0x6b, 0x64, 0x6f, 0x77, 0x6e, 0x45, 0x6e,
+	0x74, 0x72, 0x79, 0x12, 0x10, 0x0a, 0x03, 0x6b, 0x65, 0x79, 0x18, 0x01, 0x20, 0x01, 0x28, 0x09,
+	0x52, 0x03, 0x6b, 0x65, 0x79, 0x12, 0x14, 0x0a, 0x05, 0x76, 0x61, 0x6c, 0x75, 0x65, 0x18, 0x02,
+	0x20, 0x01, 0x28, 0x04, 0x52, 0x05, 0x76, 0x61, 0x6c, 0x75, 0x65, 0x3a, 0x02, 0x38, 0x01, 0x22,
+	0xbc, 0x01, 0x0a, 0x0e, 0x53, 0x74, 0x61, 0x74, 0x75, 0x73, 0x52, 0x65, 0x73, 0x70, 0x6f, 0x6e,
+	0x73, 0x65, 0x12, 0x33, 0x0a, 0x05, 0x73, 0x74, 0x61, 0x74, 0x65, 0x18, 0x01, 0x20, 0x01, 0x28,
+	0x0e, 0x32, 0x1d, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x53, 0x74, 0x61, 0x74,
+	0x75, 0x73, 0x52, 0x65, 0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65, 0x2e, 0x53, 0x74, 0x61, 0x74, 0x65,
+	0x52, 0x05, 0x73, 0x74, 0x61, 0x74, 0x65, 0x12, 0x30, 0x0a, 0x06, 0x6d, 0x65, 0x6d, 0x6f, 0x72,
+	0x79, 0x18, 0x02, 0x20, 0x01, 0x28, 0x0b, 0x32, 0x18, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e,
+	0x64, 0x2e, 0x4d, 0x65, 0x6d, 0x6f, 0x72, 0x79, 0x55, 0x73, 0x61, 0x67, 0x65, 0x44, 0x61, 0x74,
+	0x61, 0x52, 0x06, 0x6d, 0x65, 0x6d, 0x6f, 0x72, 0x79, 0x22, 0x43, 0x0a, 0x05, 0x53, 0x74, 0x61,
+	0x74, 0x65, 0x12, 0x11, 0x0a, 0x0d, 0x55, 0x4e, 0x49, 0x4e, 0x49, 0x54, 0x49, 0x41, 0x4c, 0x49,
+	0x5a, 0x45, 0x44, 0x10, 0x00, 0x12, 0x08, 0x0a, 0x04, 0x42, 0x55, 0x53, 0x59, 0x10, 0x01, 0x12,
+	0x09, 0x0a, 0x05, 0x52, 0x45, 0x41, 0x44, 0x59, 0x10, 0x02, 0x12, 0x12, 0x0a, 0x05, 0x45, 0x52,
+	0x52, 0x4f, 0x52, 0x10, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x01, 0x32, 0xf4,
+	0x04, 0x0a, 0x07, 0x42, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x12, 0x32, 0x0a, 0x06, 0x48, 0x65,
+	0x61, 0x6c, 0x74, 0x68, 0x12, 0x16, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x48,
+	0x65, 0x61, 0x6c, 0x74, 0x68, 0x4d, 0x65, 0x73, 0x73, 0x61, 0x67, 0x65, 0x1a, 0x0e, 0x2e, 0x62,
+	0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x52, 0x65, 0x70, 0x6c, 0x79, 0x22, 0x00, 0x12, 0x34,
+	0x0a, 0x07, 0x50, 0x72, 0x65, 0x64, 0x69, 0x63, 0x74, 0x12, 0x17, 0x2e, 0x62, 0x61, 0x63, 0x6b,
+	0x65, 0x6e, 0x64, 0x2e, 0x50, 0x72, 0x65, 0x64, 0x69, 0x63, 0x74, 0x4f, 0x70, 0x74, 0x69, 0x6f,
+	0x6e, 0x73, 0x1a, 0x0e, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x52, 0x65, 0x70,
+	0x6c, 0x79, 0x22, 0x00, 0x12, 0x35, 0x0a, 0x09, 0x4c, 0x6f, 0x61, 0x64, 0x4d, 0x6f, 0x64, 0x65,
+	0x6c, 0x12, 0x15, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x4d, 0x6f, 0x64, 0x65,
+	0x6c, 0x4f, 0x70, 0x74, 0x69, 0x6f, 0x6e, 0x73, 0x1a, 0x0f, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65,
+	0x6e, 0x64, 0x2e, 0x52, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x22, 0x00, 0x12, 0x3c, 0x0a, 0x0d, 0x50,
+	0x72, 0x65, 0x64, 0x69, 0x63, 0x74, 0x53, 0x74, 0x72, 0x65, 0x61, 0x6d, 0x12, 0x17, 0x2e, 0x62,
+	0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x50, 0x72, 0x65, 0x64, 0x69, 0x63, 0x74, 0x4f, 0x70,
+	0x74, 0x69, 0x6f, 0x6e, 0x73, 0x1a, 0x0e, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e,
+	0x52, 0x65, 0x70, 0x6c, 0x79, 0x22, 0x00, 0x30, 0x01, 0x12, 0x40, 0x0a, 0x09, 0x45, 0x6d, 0x62,
+	0x65, 0x64, 0x64, 0x69, 0x6e, 0x67, 0x12, 0x17, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64,
+	0x2e, 0x50, 0x72, 0x65, 0x64, 0x69, 0x63, 0x74, 0x4f, 0x70, 0x74, 0x69, 0x6f, 0x6e, 0x73, 0x1a,
+	0x18, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x45, 0x6d, 0x62, 0x65, 0x64, 0x64,
+	0x69, 0x6e, 0x67, 0x52, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x22, 0x00, 0x12, 0x41, 0x0a, 0x0d, 0x47,
+	0x65, 0x6e, 0x65, 0x72, 0x61, 0x74, 0x65, 0x49, 0x6d, 0x61, 0x67, 0x65, 0x12, 0x1d, 0x2e, 0x62,
+	0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x47, 0x65, 0x6e, 0x65, 0x72, 0x61, 0x74, 0x65, 0x49,
+	0x6d, 0x61, 0x67, 0x65, 0x52, 0x65, 0x71, 0x75, 0x65, 0x73, 0x74, 0x1a, 0x0f, 0x2e, 0x62, 0x61,
+	0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x52, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x22, 0x00, 0x12, 0x4d,
+	0x0a, 0x12, 0x41, 0x75, 0x64, 0x69, 0x6f, 0x54, 0x72, 0x61, 0x6e, 0x73, 0x63, 0x72, 0x69, 0x70,
+	0x74, 0x69, 0x6f, 0x6e, 0x12, 0x1a, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x54,
+	0x72, 0x61, 0x6e, 0x73, 0x63, 0x72, 0x69, 0x70, 0x74, 0x52, 0x65, 0x71, 0x75, 0x65, 0x73, 0x74,
+	0x1a, 0x19, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x54, 0x72, 0x61, 0x6e, 0x73,
+	0x63, 0x72, 0x69, 0x70, 0x74, 0x52, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x22, 0x00, 0x12, 0x2d, 0x0a,
+	0x03, 0x54, 0x54, 0x53, 0x12, 0x13, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x54,
+	0x54, 0x53, 0x52, 0x65, 0x71, 0x75, 0x65, 0x73, 0x74, 0x1a, 0x0f, 0x2e, 0x62, 0x61, 0x63, 0x6b,
+	0x65, 0x6e, 0x64, 0x2e, 0x52, 0x65, 0x73, 0x75, 0x6c, 0x74, 0x22, 0x00, 0x12, 0x4a, 0x0a, 0x0e,
+	0x54, 0x6f, 0x6b, 0x65, 0x6e, 0x69, 0x7a, 0x65, 0x53, 0x74, 0x72, 0x69, 0x6e, 0x67, 0x12, 0x17,
+	0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x50, 0x72, 0x65, 0x64, 0x69, 0x63, 0x74,
+	0x4f, 0x70, 0x74, 0x69, 0x6f, 0x6e, 0x73, 0x1a, 0x1d, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e,
+	0x64, 0x2e, 0x54, 0x6f, 0x6b, 0x65, 0x6e, 0x69, 0x7a, 0x61, 0x74, 0x69, 0x6f, 0x6e, 0x52, 0x65,
+	0x73, 0x70, 0x6f, 0x6e, 0x73, 0x65, 0x22, 0x00, 0x12, 0x3b, 0x0a, 0x06, 0x53, 0x74, 0x61, 0x74,
+	0x75, 0x73, 0x12, 0x16, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x48, 0x65, 0x61,
+	0x6c, 0x74, 0x68, 0x4d, 0x65, 0x73, 0x73, 0x61, 0x67, 0x65, 0x1a, 0x17, 0x2e, 0x62, 0x61, 0x63,
+	0x6b, 0x65, 0x6e, 0x64, 0x2e, 0x53, 0x74, 0x61, 0x74, 0x75, 0x73, 0x52, 0x65, 0x73, 0x70, 0x6f,
+	0x6e, 0x73, 0x65, 0x22, 0x00, 0x42, 0x5a, 0x0a, 0x19, 0x69, 0x6f, 0x2e, 0x73, 0x6b, 0x79, 0x6e,
+	0x65, 0x74, 0x2e, 0x6c, 0x6f, 0x63, 0x61, 0x6c, 0x61, 0x69, 0x2e, 0x62, 0x61, 0x63, 0x6b, 0x65,
+	0x6e, 0x64, 0x42, 0x0e, 0x4c, 0x6f, 0x63, 0x61, 0x6c, 0x41, 0x49, 0x42, 0x61, 0x63, 0x6b, 0x65,
+	0x6e, 0x64, 0x50, 0x01, 0x5a, 0x2b, 0x67, 0x69, 0x74, 0x68, 0x75, 0x62, 0x2e, 0x63, 0x6f, 0x6d,
+	0x2f, 0x67, 0x6f, 0x2d, 0x73, 0x6b, 0x79, 0x6e, 0x65, 0x74, 0x2f, 0x4c, 0x6f, 0x63, 0x61, 0x6c,
+	0x41, 0x49, 0x2f, 0x70, 0x6b, 0x67, 0x2f, 0x67, 0x72, 0x70, 0x63, 0x2f, 0x70, 0x72, 0x6f, 0x74,
+	0x6f, 0x62, 0x06, 0x70, 0x72, 0x6f, 0x74, 0x6f, 0x33,
 }
 
 var (
diff --git a/pkg/grpc/proto/backend_grpc.pb.go b/pkg/grpc/proto/backend_grpc.pb.go
index d41f77a6..ef5187bc 100644
--- a/pkg/grpc/proto/backend_grpc.pb.go
+++ b/pkg/grpc/proto/backend_grpc.pb.go
@@ -1,7 +1,7 @@
 // Code generated by protoc-gen-go-grpc. DO NOT EDIT.
 // versions:
 // - protoc-gen-go-grpc v1.2.0
-// - protoc             v3.6.1
+// - protoc             v4.23.4
 // source: backend.proto
 
 package proto