Documentation
¶
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type ModelClient ¶ added in v0.44.0
type ModelClient interface {
GetModel(ctx context.Context, in *mv1.GetModelRequest, opts ...grpc.CallOption) (*mv1.Model, error)
}
ModelClient is an interface for a model client.
type NoopModelClient ¶ added in v0.44.0
type NoopModelClient struct { }
NoopModelClient is a no-op model client.
func (*NoopModelClient) GetModel ¶ added in v0.44.0
func (c *NoopModelClient) GetModel(ctx context.Context, in *mv1.GetModelRequest, opts ...grpc.CallOption) (*mv1.Model, error)
GetModel is a no-op implementation of GetModel.
type NoopRewriter ¶ added in v0.139.0
type NoopRewriter struct { }
NoopRewriter is a no-op rewriter.
func (*NoopRewriter) ProcessMessages ¶ added in v0.139.0
func (r *NoopRewriter) ProcessMessages( ctx context.Context, vstore *vsv1.VectorStore, messages []*v1.CreateChatCompletionRequest_Message, ) ([]*v1.CreateChatCompletionRequest_Message, error)
ProcessMessages is a no-op implementation of ProcessMessages.
type NoopVectorStoreClient ¶ added in v0.142.0
type NoopVectorStoreClient struct { }
NoopVectorStoreClient is a no-op vector store client.
func (*NoopVectorStoreClient) GetVectorStoreByName ¶ added in v0.142.0
func (c *NoopVectorStoreClient) GetVectorStoreByName(ctx context.Context, req *vsv1.GetVectorStoreByNameRequest, opts ...grpc.CallOption) (*vsv1.VectorStore, error)
GetVectorStoreByName is a no-op implementation of GetVectorStoreByName.
type Rewriter ¶ added in v0.139.0
type Rewriter interface { ProcessMessages( ctx context.Context, vstore *vsv1.VectorStore, messages []*v1.CreateChatCompletionRequest_Message, ) ([]*v1.CreateChatCompletionRequest_Message, error) }
Rewriter is an interface for rag.
type S ¶
type S struct { v1.UnimplementedChatServiceServer // contains filtered or unexported fields }
S is a server.
func New ¶
func New( m monitoring.MetricsMonitoring, modelClient ModelClient, vsClient VectorStoreClient, r Rewriter, taskQueue *infprocessor.TaskQueue, ) *S
New creates a server.
func (*S) CreateChatCompletion ¶
func (s *S) CreateChatCompletion( w http.ResponseWriter, req *http.Request, pathParams map[string]string, )
CreateChatCompletion creates a chat completion.
type VectorStoreClient ¶ added in v0.142.0
type VectorStoreClient interface {
GetVectorStoreByName(ctx context.Context, req *vsv1.GetVectorStoreByNameRequest, opts ...grpc.CallOption) (*vsv1.VectorStore, error)
}
VectorStoreClient is an interface for a vector store client.
type WS ¶ added in v0.112.0
type WS struct { v1.UnimplementedInferenceWorkerServiceServer // contains filtered or unexported fields }
WS is a server for worker services.
func NewWorkerServiceServer ¶ added in v0.112.0
func NewWorkerServiceServer(infProcessor *infprocessor.P) *WS
NewWorkerServiceServer creates a new worker service server.
func (*WS) ProcessTasks ¶ added in v0.115.0
func (ws *WS) ProcessTasks(srv v1.InferenceWorkerService_ProcessTasksServer) error
ProcessTasks processes tasks.