update

components/llm/deepinfra_api.py

@@ -328,7 +328,6 @@ class DeepInfraApi(LlmApi):
         Yields:
             str: Токены ответа LLM.
         """
-        print(request.history)
         timeout = httpx.Timeout(connect=30.0, read=None, pool=None, write=None, timeout=None)
         attempt = 0
         

routes/llm.py

@@ -2,6 +2,7 @@ import json
 import logging
 import os
 from typing import Annotated, AsyncGenerator, List, Optional
+import asyncio
 
 from fastapi import APIRouter, Depends, HTTPException
 from fastapi.responses import StreamingResponse
@@ -124,6 +125,7 @@ def collapse_history_to_first_message(chat_request: ChatRequest) -> ChatRequest:
         searchResults=''
     )
     return ChatRequest(history=[new_message])
+
         
 async def sse_generator(request: ChatRequest, llm_api: DeepInfraApi, system_prompt: str, 
                         predict_params: LlmPredictParams,
@@ -166,7 +168,16 @@ async def sse_generator(request: ChatRequest, llm_api: DeepInfraApi, system_prom
             previous_entities, chunk_ids, scores = entity_service.search_similar(qe_result.search_query, 
                                                                                  dataset.id, previous_entities)
             text_chunks = entity_service.build_text(chunk_ids, scores)
-            all_text_chunks = [text_chunks] + [entity_service.build_text(entities) for entities in previous_entities]
+            
+            
+            async def build_text_async(entities):
+                return await asyncio.to_thread(entity_service.build_text, entities)
+
+            # all_text_chunks = [text_chunks] + [entity_service.build_text(entities) for entities in previous_entities]
+            tasks = [build_text_async(entities) for entities in previous_entities]
+            built_texts = await asyncio.gather(*tasks)
+            
+            all_text_chunks = [text_chunks] + built_texts
             all_entities = [chunk_ids] + previous_entities
             
             search_results_event = {