rag-pipeline-ru-v0

Question it answers

Where does a Russian-language RAG pipeline actually lose quality, component by component?

What is measured

Per-component evaluation: OCR CER/WER → splitter boundary F1 → Recall@k/nDCG → reranker delta → generation support/abstention/citations. Internal until 6 release gates pass.

Claims this workload cannot support

Any commercial use before release gates; end-to-end quality claims from a single component metric.

Identity discipline

Every released workload revision freezes its item hashes, tokenizer revision, load model, cache policy and quality gates. Results reference the exact revision; changing any identity field creates a new revision.