WueRAG_2025_08_22 — Report Generation Task

Document collection: ['English subset']
Machine translation of documents: ['None']
Write a short description of your retrieval process: The retrieval process starts by reformulating the user query for better search relevance. Depending on configuration, the system either queries a remote RAGTIME API or searches local indices (for the last submission the RAGTIME API was used with subset set to "eng"). Retrieved documents are converted into a temporary index and processed via an embedding-based and BM25 retriever, in an hybrid setting.
Write a short description of your generation process: The generation process uses the retrieved nodes to produce a final answer straight forward. A response synthesizer applies a prompt template instructing the model to answer strictly from the provided sources and cite them inline. The synthesizer also refines responses using a citation-aware template to ensure factual accuracy. If the response does not fit the problem_statement (in terms of content or length limits,...), the process gets repeated.
Which LLM(s) where used by your system?: sentence-transformers/all-MiniLM-L6-v2 for embedding and gpt4o for generation
Open repository link: -
Assessing priority: 5