5514356c53e976f014f16eba1ded939a4a674d77
- Added automatic Ollama embedding model discovery in LlamaServerManager (bge, nomic, embed, minilm) - Implemented dynamic routing in RagService for embeddings (Ollama 11434 vs local llama-server 8081) - Added embedding model consistency check with embed_model.tag to prevent Lucene vector dimension mismatches - Added Lucene store maintenance functions: getLuceneStats, resetLuceneStore, and automatic 50MB retention cleanup - Added real-time Lucene status display and one-click '벡터 DB 초기화' button to TradeConfigPanel - Verified with compileKotlin and full project build
분석용 llm
https://huggingface.co/MLP-KTLim/llama-3-Korean-Bllossom-8B-gguf-Q4_K_M
https://gofile.me/6z2kw/ypnQYrMl3 분석 모델
뉴스용 임베디드 llm
https://gofile.me/6z2kw/xMbFVg1f1 임베딩 모델
bge-m3-q4_k_m.gguf
Languages
Kotlin
100%