o پiT@sdZddlmZddlmZddlmZddlmZ ddl m Z e ejjddd d d e Zdd lmZdd lmZd dlmZedZeZe eee edeejstde_ ded_ ded_!ded_ ded_!ded_"ded_#ded_"ded_#ded_"ded_#ded_"ded_#d ed!_"d"ed!_#d#ed$_"d%ed$_#d&ed'_"d(ed'_#d)ed_"d*ed_#d+ed,_"d-ed,_#d.ed/_"d0ed/_#d1ed2_"d3ed2_#d4ed5_"d6ed5_#d7ed8_"d9ed8_#d:ed;_"d_"d?ed>_#d@edA_"dBedA_#dCedD_"dEedD_#dFedG_"dHedG_#dIedJ_"dKedJ_#dLedM_"dNedM_#dOedP_"dQedP_#dRedS_"dTedS_#dUedV_"dWedV_#dXedY_"dZedY_#d[ed\_"d]ed\_#d^ed__"d`ed__#daedb_"dcedb_#ddede_"dfede_#dgedh_"diedh_#djedk_"dledk_#dmedn_"doedn_#dpedq_"dredq_#dsedt_"duedt_#dvedw_"dxedw_#dyedz_"d{edz_#d|ed}_"d~ed}_#ded_"ded_#ded_"ded_#ded_"ded_#ded_"ded_#ded_"ded_#ded_"ded_#ded_"ded_#ded_"ded_#ded_"ded_#ded_"ded_#ded_"ded_#ded_"ded_#ded_"ded_#ded_"ded_#ded_"ded_#dSdS)zGenerated protocol buffer code.) descriptor)descriptor_pool)runtime_version)symbol_database)builderzsglang_scheduler.proto) timestamp_pb2) struct_pb2) common_pb2s" sglang_scheduler.protosglang.grpc.schedulergoogle/protobuf/timestamp.protogoogle/protobuf/struct.proto common.proto" SamplingParams temperature ( top_p ( top_k ( min_p ( frequency_penalty ( presence_penalty ( repetition_penalty ( max_new_tokens ( H stop (  stop_token_ids (  skip_special_tokens (% spaces_between_special_tokens ( regex ( H json_schema ( H ebnf_grammar ( H structural_tag ( H n (  min_new_tokens (  ignore_eos ( no_stop_trim ( stream_interval (HH logit_bias ( 24.sglang.grpc.scheduler.SamplingParams.LogitBiasEntry. custom_params ( 2.google.protobuf.Struct0 LogitBiasEntry key (  value (:8B constraintB _max_new_tokensB _stream_interval"] DisaggregatedParams bootstrap_host (  bootstrap_port ( bootstrap_room (" GenerateRequest request_id ( 8 tokenized ( 2%.sglang.grpc.scheduler.TokenizedInput: mm_inputs ( 2'.sglang.grpc.scheduler.MultimodalInputs> sampling_params ( 2%.sglang.grpc.scheduler.SamplingParams return_logprob ( logprob_start_len ( top_logprobs_num ( token_ids_logprob (  return_hidden_states (H disaggregated_params ( 2*.sglang.grpc.scheduler.DisaggregatedParams custom_logit_processor ( - timestamp ( 2.google.protobuf.Timestamp log_metrics ( input_embeds ( lora_id (  data_parallel_rank ( stream (": TokenizedInput original_text (  input_ids ( "8 TensorData data (  shape (  dtype ( " MultimodalInputs image_urls (  video_urls (  audio_urls ( 7 pixel_values ( 2!.sglang.grpc.scheduler.TensorData image_data (  video_data (  audio_data (  modalities ( a model_specific_tensors ( 2A.sglang.grpc.scheduler.MultimodalInputs.ModelSpecificTensorsEntry im_token_id ( H@ mm_placeholders ( 2'.sglang.grpc.scheduler.PlaceholderRange^ ModelSpecificTensorsEntry key ( 0 value ( 2!.sglang.grpc.scheduler.TensorData:8B _im_token_id"2 PlaceholderRange offset (  length ( " GenerateResponse request_id ( ; chunk ( 2*.sglang.grpc.scheduler.GenerateStreamChunkH; complete ( 2'.sglang.grpc.scheduler.GenerateCompleteH5 error ( 2$.sglang.grpc.scheduler.GenerateErrorHB response" GenerateStreamChunk token_ids (  prompt_tokens (  completion_tokens (  cached_tokens ( > output_logprobs ( 2%.sglang.grpc.scheduler.OutputLogProbs hidden_states (< input_logprobs ( 2$.sglang.grpc.scheduler.InputLogProbs index ( " GenerateComplete output_ids (  finish_reason (  prompt_tokens (  completion_tokens (  cached_tokens ( > output_logprobs ( 2%.sglang.grpc.scheduler.OutputLogProbs> all_hidden_states ( 2#.sglang.grpc.scheduler.HiddenStates matched_token_id ( H matched_stop_str ( H< input_logprobs ( 2$.sglang.grpc.scheduler.InputLogProbs index ( B matched_stop"K GenerateError message (  http_status_code (  details ( "u OutputLogProbs token_logprobs ( token_ids ( 8 top_logprobs ( 2".sglang.grpc.scheduler.TopLogProbs" InputLogProbs@ token_logprobs ( 2(.sglang.grpc.scheduler.InputTokenLogProb token_ids ( 8 top_logprobs ( 2".sglang.grpc.scheduler.TopLogProbs"1 InputTokenLogProb value (HB _value"0 TopLogProbs values ( token_ids ( "? HiddenStates values ( layer ( position (" EmbedRequest request_id ( 8 tokenized ( 2%.sglang.grpc.scheduler.TokenizedInput: mm_inputs ( 2'.sglang.grpc.scheduler.MultimodalInputs> sampling_params ( 2%.sglang.grpc.scheduler.SamplingParams log_metrics ( token_type_ids ( data_parallel_rank ( is_cross_encoder ( texts ( " EmbedResponse request_id ( 8 complete ( 2$.sglang.grpc.scheduler.EmbedCompleteH2 error ( 2!.sglang.grpc.scheduler.EmbedErrorHB response" EmbedComplete embedding ( prompt_tokens (  cached_tokens (  embedding_dim ( : batch_embeddings ( 2 .sglang.grpc.scheduler.Embedding"* Embedding values ( index ("< EmbedError message (  code (  details ( " HealthCheckRequest"7 HealthCheckResponse healthy ( message ( "2 AbortRequest request_id (  reason ( "1 AbortResponse success ( message ( "I LoadLoRARequest adapter_id (  adapter_path (  rank ("H LoadLoRAResponse success ( adapter_id (  message ( "' UnloadLoRARequest adapter_id ( "6 UnloadLoRAResponse success ( message ( "w UpdateWeightsRequest disk_path ( H tensor_data ( H remote_url ( H weight_name ( B source"9 UpdateWeightsResponse success ( message ( "- GetInternalStateRequest state_keys ( "B GetInternalStateResponse& state ( 2.google.protobuf.Struct"A SetInternalStateRequest& state ( 2.google.protobuf.Struct"< SetInternalStateResponse success ( message ( " GetModelInfoRequest" GetModelInfoResponse model_path (  tokenizer_path (  is_generation (! preferred_sampling_params (  weight_version (  served_model_name (  max_context_length ( vocab_size ( supports_vision ( model_type (  eos_token_ids ( pad_token_id ( bos_token_id ( max_req_input_len ( architectures (  id2label_json (  num_labels (" GetServerInfoRequest" GetServerInfoResponse, server_args ( 2.google.protobuf.Struct/ scheduler_info ( 2.google.protobuf.Struct active_requests ( is_paused ( last_receive_timestamp ( uptime_seconds ( sglang_version (  server_type ( . start_time ( 2.google.protobuf.Timestamp"D GetLoadsRequest dp_rank (H include ( B _dp_rank" GetLoadsResponse timestamp (  version (  dp_rank_count (3 loads ( 2$.sglang.grpc.scheduler.SchedulerLoad: aggregate ( 2'.sglang.grpc.scheduler.AggregateMetrics" SchedulerLoad dp_rank ( num_running_reqs ( num_waiting_reqs ( num_total_reqs ( num_used_tokens ( max_total_num_tokens ( token_usage ( gen_throughput ( cache_hit_rate ( utilization ( max_running_requests (9 memory ( 2$.sglang.grpc.scheduler.MemoryMetricsHC speculative ( 2).sglang.grpc.scheduler.SpeculativeMetricsH5 lora ( 2".sglang.grpc.scheduler.LoRAMetricsHI disaggregation ( 2,.sglang.grpc.scheduler.DisaggregationMetricsH8 queues ( 2#.sglang.grpc.scheduler.QueueMetricsHB _memoryB _speculativeB _loraB _disaggregationB _queues"a MemoryMetrics weight_gb ( kv_cache_gb ( graph_gb ( token_capacity ("@ SpeculativeMetrics accept_length ( accept_rate ("K LoRAMetrics slots_used ( slots_total ( utilization (" DisaggregationMetrics mode ( # prefill_prealloc_queue_reqs (# prefill_inflight_queue_reqs (" decode_prealloc_queue_reqs (" decode_transfer_queue_reqs (# decode_retracted_queue_reqs ( kv_transfer_speed_gb_s ( kv_transfer_latency_ms ("S QueueMetrics waiting ( grammar ( paused ( retracted (" AggregateMetrics total_running_reqs ( total_waiting_reqs ( total_reqs ( avg_token_usage ( avg_throughput ( avg_utilization (2 SglangScheduler] Generate&.sglang.grpc.scheduler.GenerateRequest'.sglang.grpc.scheduler.GenerateResponse0R Embed#.sglang.grpc.scheduler.EmbedRequest$.sglang.grpc.scheduler.EmbedResponsed HealthCheck).sglang.grpc.scheduler.HealthCheckRequest*.sglang.grpc.scheduler.HealthCheckResponseR Abort#.sglang.grpc.scheduler.AbortRequest$.sglang.grpc.scheduler.AbortResponseg GetModelInfo*.sglang.grpc.scheduler.GetModelInfoRequest+.sglang.grpc.scheduler.GetModelInfoResponsej GetServerInfo+.sglang.grpc.scheduler.GetServerInfoRequest,.sglang.grpc.scheduler.GetServerInfoResponse[ GetLoads&.sglang.grpc.scheduler.GetLoadsRequest'.sglang.grpc.scheduler.GetLoadsResponseZ GetTokenizer$.smg.grpc.common.GetTokenizerRequest".smg.grpc.common.GetTokenizerChunk0_ SubscribeKvEvents).smg.grpc.common.SubscribeKvEventsRequest.smg.grpc.common.KvEventBatch0bproto3sglang_scheduler_pb2N_SAMPLINGPARAMS_LOGITBIASENTRYs8+_MULTIMODALINPUTS_MODELSPECIFICTENSORSENTRY_SAMPLINGPARAMSiOiiiQ_DISAGGREGATEDPARAMSii_GENERATEREQUESTii_TOKENIZEDINPUTiOiQ _TENSORDATAii_MULTIMODALINPUTSii$ii_PLACEHOLDERRANGEii_GENERATERESPONSEi i _GENERATESTREAMCHUNKi i _GENERATECOMPLETEib id _GENERATEERRORi i _OUTPUTLOGPROBSi& i) _INPUTLOGPROBSi i _INPUTTOKENLOGPROBi i _TOPLOGPROBSi,i. _HIDDENSTATESimip _EMBEDREQUESTii_EMBEDRESPONSEiZi]_EMBEDCOMPLETEii _EMBEDDINGi,i. _EMBEDERRORijil_HEALTHCHECKREQUESTii_HEALTHCHECKRESPONSEii _ABORTREQUESTii_ABORTRESPONSEi i"_LOADLORAREQUESTikim_LOADLORARESPONSEii_UNLOADLORAREQUESTii_UNLOADLORARESPONSEii_UPDATEWEIGHTSREQUESTii_UPDATEWEIGHTSRESPONSEii_GETINTERNALSTATEREQUESTii_GETINTERNALSTATERESPONSEi=i?_SETINTERNALSTATEREQUESTii_SETINTERNALSTATERESPONSEii_GETMODELINFOREQUESTii_GETMODELINFORESPONSEii_GETSERVERINFOREQUESTii_GETSERVERINFORESPONSEii_GETLOADSREQUESTii_GETLOADSRESPONSEii_SCHEDULERLOADiyi{_MEMORYMETRICSii_SPECULATIVEMETRICSii  _LORAMETRICSikin_DISAGGREGATIONMETRICSii _QUEUEMETRICSii_AGGREGATEMETRICSii_SGLANGSCHEDULERi!)$__doc__google.protobufr _descriptorr_descriptor_poolr_runtime_versionr_symbol_databasegoogle.protobuf.internalr_builderValidateProtobufRuntimeVersionDomainPUBLICDefault_sym_dbr &google_dot_protobuf_dot_timestamp__pb2r #google_dot_protobuf_dot_struct__pb2r r common__pb2AddSerializedFile DESCRIPTORglobals_globalsBuildMessageAndEnumDescriptorsBuildTopDescriptorsAndMessages_USE_C_DESCRIPTORS_loaded_options_serialized_options_serialized_start_serialized_endr^r^a/home/ubuntu/.local/lib/python3.10/site-packages/smg_grpc_proto/generated/sglang_scheduler_pb2.pys