Non English, Chinese documents processing Entity generation, and Ebilingual chat

[kparchure]

Does the application support uploading 80+ around 10-300 KB each html/PDF Chinese documents? Somehow the chat interface does not seem to reference all documents for Q&A.

Any pointers for Graph Enhancement/Entity Extraction, Additional Instruction to meaningfully extract entities for a particular domain (since the instructions are in English and documents in Chinese). Tried preprocessing instruction in English but process seem to get stuck (file fails to process).

without any preprocessing instruction or Entity Extraction settings the documents process and we're able to chat in English, but the answers are not always accurate, and seems like the entire set of documents/context is not being used.

we're running LLM builder, neo4j db, and both qwen 2.5, QwQ-32B-AWQ locally