feat: Add Get Token Metrics to GRPC server (#3687)

* Add Get Token Metrics to GRPC server

Signed-off-by: Siddharth More <siddimore@gmail.com>

* Expose LocalAI endpoint

Signed-off-by: Siddharth More <siddimore@gmail.com>

---------

Signed-off-by: Siddharth More <siddimore@gmail.com>
This commit is contained in:
siddimore 2024-10-01 05:41:20 -07:00 committed by GitHub
parent 139209353f
commit f84b55d1ef
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
8 changed files with 180 additions and 0 deletions

View file

@ -26,6 +26,19 @@ service Backend {
rpc StoresFind(StoresFindOptions) returns (StoresFindResult) {}
rpc Rerank(RerankRequest) returns (RerankResult) {}
rpc GetMetrics(MetricsRequest) returns (MetricsResponse);
}
// Define the empty request
message MetricsRequest {}
message MetricsResponse {
int32 slot_id = 1;
string prompt_json_for_slot = 2; // Stores the prompt as a JSON string.
float tokens_per_second = 3;
int32 tokens_generated = 4;
int32 prompt_tokens_processed = 5;
}
message RerankRequest {