LocalAI/pkg/grpc/proto/backend.proto
Ettore Di Giacinto a28ab18987
feat(vllm): Allow to set quantization (#1094)
This particularly useful to set AWQ

**Description**

Follow up of #1015 

**Notes for Reviewers**


**[Signed
commits](../CONTRIBUTING.md#signing-off-on-commits-developer-certificate-of-origin)**
- [ ] Yes, I signed my commits.
 

<!--
Thank you for contributing to LocalAI! 

Contributing Conventions:

1. Include descriptive PR titles with [<component-name>] prepended.
2. Build and test your changes before submitting a PR. 
3. Sign your commits

By following the community's contribution conventions upfront, the
review process will
be accelerated and your PR merged more quickly.
-->

---------

Signed-off-by: Ettore Di Giacinto <mudler@localai.io>
2023-09-22 15:52:38 +02:00

197 lines
4.1 KiB
Protocol Buffer

syntax = "proto3";
option go_package = "github.com/go-skynet/LocalAI/pkg/grpc/proto";
option java_multiple_files = true;
option java_package = "io.skynet.localai.backend";
option java_outer_classname = "LocalAIBackend";
package backend;
service Backend {
rpc Health(HealthMessage) returns (Reply) {}
rpc Predict(PredictOptions) returns (Reply) {}
rpc LoadModel(ModelOptions) returns (Result) {}
rpc PredictStream(PredictOptions) returns (stream Reply) {}
rpc Embedding(PredictOptions) returns (EmbeddingResult) {}
rpc GenerateImage(GenerateImageRequest) returns (Result) {}
rpc AudioTranscription(TranscriptRequest) returns (TranscriptResult) {}
rpc TTS(TTSRequest) returns (Result) {}
rpc TokenizeString(PredictOptions) returns (TokenizationResponse) {}
rpc Status(HealthMessage) returns (StatusResponse) {}
}
message HealthMessage {}
// The request message containing the user's name.
message PredictOptions {
string Prompt = 1;
int32 Seed = 2;
int32 Threads = 3;
int32 Tokens = 4;
int32 TopK = 5;
int32 Repeat = 6;
int32 Batch = 7;
int32 NKeep = 8;
float Temperature = 9;
float Penalty = 10;
bool F16KV = 11;
bool DebugMode = 12;
repeated string StopPrompts = 13;
bool IgnoreEOS = 14;
float TailFreeSamplingZ = 15;
float TypicalP = 16;
float FrequencyPenalty = 17;
float PresencePenalty = 18;
int32 Mirostat = 19;
float MirostatETA = 20;
float MirostatTAU = 21;
bool PenalizeNL = 22;
string LogitBias = 23;
bool MLock = 25;
bool MMap = 26;
bool PromptCacheAll = 27;
bool PromptCacheRO = 28;
string Grammar = 29;
string MainGPU = 30;
string TensorSplit = 31;
float TopP = 32;
string PromptCachePath = 33;
bool Debug = 34;
repeated int32 EmbeddingTokens = 35;
string Embeddings = 36;
float RopeFreqBase = 37;
float RopeFreqScale = 38;
float NegativePromptScale = 39;
string NegativePrompt = 40;
int32 NDraft = 41;
}
// The response message containing the result
message Reply {
bytes message = 1;
}
message ModelOptions {
string Model = 1;
int32 ContextSize = 2;
int32 Seed = 3;
int32 NBatch = 4;
bool F16Memory = 5;
bool MLock = 6;
bool MMap = 7;
bool VocabOnly = 8;
bool LowVRAM = 9;
bool Embeddings = 10;
bool NUMA = 11;
int32 NGPULayers = 12;
string MainGPU = 13;
string TensorSplit = 14;
int32 Threads = 15;
string LibrarySearchPath = 16;
float RopeFreqBase = 17;
float RopeFreqScale = 18;
float RMSNormEps = 19;
int32 NGQA = 20;
string ModelFile = 21;
// AutoGPTQ
string Device = 22;
bool UseTriton = 23;
string ModelBaseName = 24;
bool UseFastTokenizer = 25;
// Diffusers
string PipelineType = 26;
string SchedulerType = 27;
bool CUDA = 28;
float CFGScale = 29;
bool IMG2IMG = 30;
string CLIPModel = 31;
string CLIPSubfolder = 32;
int32 CLIPSkip = 33;
// RWKV
string Tokenizer = 34;
// LLM (llama.cpp)
string LoraBase = 35;
string LoraAdapter = 36;
bool NoMulMatQ = 37;
string DraftModel = 39;
string AudioPath = 38;
// vllm
string Quantization = 40;
}
message Result {
string message = 1;
bool success = 2;
}
message EmbeddingResult {
repeated float embeddings = 1;
}
message TranscriptRequest {
string dst = 2;
string language = 3;
uint32 threads = 4;
}
message TranscriptResult {
repeated TranscriptSegment segments = 1;
string text = 2;
}
message TranscriptSegment {
int32 id = 1;
int64 start = 2;
int64 end = 3;
string text = 4;
repeated int32 tokens = 5;
}
message GenerateImageRequest {
int32 height = 1;
int32 width = 2;
int32 mode = 3;
int32 step = 4;
int32 seed = 5;
string positive_prompt = 6;
string negative_prompt = 7;
string dst = 8;
string src = 9;
// Diffusers
string EnableParameters = 10;
int32 CLIPSkip = 11;
}
message TTSRequest {
string text = 1;
string model = 2;
string dst = 3;
}
message TokenizationResponse {
int32 length = 1;
repeated int32 tokens = 2;
}
message MemoryUsageData {
uint64 total = 1;
map<string, uint64> breakdown = 2;
}
message StatusResponse {
enum State {
UNINITIALIZED = 0;
BUSY = 1;
READY = 2;
ERROR = -1;
}
State state = 1;
MemoryUsageData memory = 2;
}