How do I adjust LightRAG's working language?（help!）

[PolleyIU]

As shown in the figure, I have set the document processing output language to Chinese in the .env configuration file.

However, when I uploaded a Chinese classic story, LightRAG extracted a lot of English entities and relationships. This does not meet my expectations and usage requirements, and I need your help. The LLM I use is deepseek-v3, and the embedding model is bge-m3。