Retrieval-Augmented Generation Geometry, Reciprocal Rank Fusion and Multi-Agent Error Cascades: chapter answers
Retrieval-augmented generation (RAG) and multi-agent pipelines can be split into three links: retrieval, assembly and the chain. Each has its popular intuition — higher dimension and more chunks are better; a longer context is better; more agents are better. This book writes each intuition as a runnable model and finds where it holds and where it fails. All data are synthetic, generated by the book's programs; the numbers show mechanisms and orders of magnitude and are not measurements of any product.
DOI 10.5281/zenodo.24090170 · Read the monograph · 中文版
- Chapter 1 · High-Dimensional Geometry: Concentration of Measure, Effective Dimension and Hub Chunks
Chapter 1 uses concentration of measure, effective dimension and hubness to show that the larger the corpus, the higher the top-1 error (from 0.0730 to 0.5030 at 256 dimensions); that with 768 nominal dimensions and a p… - Chapter 2 · Lexical and Dense Retrieval: Their Blind Spots and the Smoothing Constant of Reciprocal Rank Fusion
Chapter 2 shows that dense retrieval cannot recognise identifiers and lexical retrieval cannot recognise paraphrases; fusing complete rankings leaves both query types at a recall@1 of 0.055, and fusing short lists leave… - Chapter 3 · The Context Window: Position Decay, Evidence Dilution and the Best Number of Chunks
Chapter 3 uses a stylised reader model to show "lost in the middle" and evidence dilution: a gold chunk in the middle is answered correctly only about a third as often as at either end (0.098 against 0.288), and the bes… - Chapter 4 · Multi-Agent Pipelines: Error Cascades and the Verifier-Gated Absorbing Markov Chain
Chapter 4 confirms that 0.92 to the tenth power is 0.4344, shows that three-way voting helps only while errors are independent, and writes verifier gating as an absorbing Markov chain: a deterministic, independent check… - Chapter 5 · Verifier Independence and the Cost of Retries
Chapter 5 shows that a verifier's value lies in independence: if hard steps stay hard even a deterministic check reaches only 0.886, and a judge sharing the generator's blind spot only 0.6950; reliability is paid for in… - Chapter 6 · The Statistics of Evaluation Itself: Sample Size, Judge Bias and the Winner's Curse
Chapter 6 shows that a 3-point improvement needs about three thousand queries to be detected reliably (power 0.989 at 3,000) and that picking the best of 256 equally good configurations inflates it by more than seven po… - Chapter 7 · The Routing of the *Qilue* and Han Fei's *Canyan*: Two Ancient Methods of Retrieval and Verification
Chapter 7 reads the Qilue as routing by category (routing must be more than 0.656 accurate to beat searching everything) and the Han Feizi's canyan as checking against several sources (sources sharing an origin are wort… - Chapter 8 · A Five-Step Check and a Pre-Registered Offline Evaluation
Chapter 8 gives a five-step check and a pre-registered offline evaluation, and shows as it fell a wrong conclusion reached through bad luck, with its error rate (0.045).
Illustrative model calculation, not a live quote