hyungi/tk-factory-services
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: AI 서비스 MLX 듀얼 백엔드 및 모델 최적화

Browse Source 
- MLX(맥미니 27B) 우선 → Ollama(조립컴 9B) fallback 구조
- pydantic-settings 기반 config 전환
- health check에 MLX 상태 추가
- 텍스트 모델 qwen3:8b → qwen3.5:9b-q8_0 변경

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
Hyungi Ahn
2026-03-06 23:17:50 +09:00
parent cad662473b
commit 2f7e083db0
14 changed files with 231 additions and 140 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
ai-service/services/rag_service.py
View File

@@ -1,11 +1,7 @@
from services.ollama_client import ollama_client
from services.embedding_service import search_similar_by_text, build_document_text
from services.db_client import get_issue_by_id
def _load_prompt(path: str) -> str:
with open(path, "r", encoding="utf-8") as f:
return f.read()
from services.utils import load_prompt
def _format_retrieved_issues(results: list[dict]) -> str:
@@ -55,7 +51,7 @@ async def rag_suggest_solution(issue_id: int) -> dict:
break
context = _format_retrieved_issues(similar)
template = _load_prompt("prompts/rag_suggest_solution.txt")
template = load_prompt("prompts/rag_suggest_solution.txt")
prompt = template.format(
description=issue.get("description", ""),
detail_notes=issue.get("detail_notes", ""),
@@ -87,7 +83,7 @@ async def rag_ask(question: str, project_id: int = None) -> dict:
)
context = _format_retrieved_issues(results)
template = _load_prompt("prompts/rag_qa.txt")
template = load_prompt("prompts/rag_qa.txt")
prompt = template.format(
question=question,
retrieved_cases=context,
@@ -113,7 +109,7 @@ async def rag_analyze_pattern(description: str, n_results: int = 10) -> dict:
results = await search_similar_by_text(description, n_results=n_results)
context = _format_retrieved_issues(results)
template = _load_prompt("prompts/rag_pattern.txt")
template = load_prompt("prompts/rag_pattern.txt")
prompt = template.format(
description=description,
retrieved_cases=context,
@@ -142,7 +138,7 @@ async def rag_classify_with_context(description: str, detail_notes: str = "") ->
similar = await search_similar_by_text(query, n_results=5)
context = _format_retrieved_issues(similar)
template = _load_prompt("prompts/rag_classify.txt")
template = load_prompt("prompts/rag_classify.txt")
prompt = template.format(
description=description,
detail_notes=detail_notes,