Multiagent LLM Debate: Real Accuracy Gains, Uncontrolled Compute, and Collective Delusion
A close reading of Du et al.'s ICML 2024 multiagent debate paper — which reports 14.8-point accuracy gains on arithmetic — alongside 2025 rebuttals showing equal-budget single agents match debate performance, and an analysis of why Collective Delusion (65% of debate failures) poses specific risks for AI-assisted ledger commits.
LLMs Are Not Useful for Time Series Forecasting: What NeurIPS 2024 Means for Finance AI
A NeurIPS 2024 Spotlight paper ablates three LLM-based time series forecasting methods — OneFitsAll, Time-LLM, and CALF — and finds that removing the language model improves accuracy in most cases, with up to a 1,383× training speedup. For finance AI applications like Beancount balance prediction, lightweight purpose-built models consistently beat repurposed LLMs.
Fine-Tuning vs. RAG: Why Retrieval Wins for Injecting New Knowledge into LLMs
Empirical comparison of RAG vs. unsupervised fine-tuning across 7B-parameter LLMs shows RAG achieves 0.875+ accuracy on post-cutoff facts while fine-tuning plateaus at 0.504 — with direct implications for Beancount agent design and any system requiring frequent knowledge updates.
IRCoT: Interleaving Retrieval with Chain-of-Thought for Multi-Step QA
IRCoT interleaves BM25 retrieval with each step of a chain-of-thought reasoning loop, achieving +11.3 retrieval recall and +7.1 F1 on HotpotQA over one-step RAG — and shows a 3B model can beat GPT-3 175B when retrieval strategy is right.
FLARE: Active Retrieval Augmented Generation
FLARE (EMNLP 2023) improves on standard RAG by triggering retrieval mid-generation using token-probability confidence thresholds, reaching 51.0 EM on 2WikiMultihopQA versus 39.4 for single-retrieval — but calibration failures in instruction-tuned chat models limit its reliability for production finance agents.
Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
Lewis et al.'s NeurIPS 2020 paper introduced the hybrid RAG architecture—a BART-large generator paired with a FAISS-indexed retriever over 21 million Wikipedia passages—achieving 44.5 EM on Natural Questions and establishing the parametric/non-parametric split that now underlies most production AI systems. This review covers RAG-Sequence vs. RAG-Token trade-offs, the retrieval collapse failure mode, and what stale indexes mean for financial AI built on append-only Beancount ledgers.
FinQA: The Benchmark Measuring AI Numerical Reasoning on Financial Reports
FinQA (EMNLP 2021) built 8,281 QA pairs from S&P 500 earnings reports requiring multi-step arithmetic programs. Neural models scored 61% at release versus 91% for human experts; accuracy collapses to 22% on three-or-more-step programs. The failure modes — domain constants, cross-modality grounding, chain length — map directly to the challenges Beancount agents face today.
DSPy: Replacing Brittle Prompt Engineering with Compiled LLM Pipelines
DSPy replaces hand-crafted prompt strings with declarative signatures and a metric-driven compiler—boosting Llama2-13b from 9.4% to 46.9% on GSM8K math reasoning and offering a more maintainable path for production finance AI pipelines.
LATS: Language Agent Tree Search — 추론, 행동, 계획을 하나의 프레임워크로 통합
LATS(Language Agent Tree Search, ICML 2024)는 ReAct, Tree of Thoughts, Reflexion을 단일 MCTS 프레임워크로 통합하여 GPT-4와 함께 HumanEval에서 92.7%의 pass@1을 달성했습니다. Git 기반의 Beancount 장부의 경우, 운영 환경에서 LATS를 제한하는 상태 복원 요구 사항을 아주 쉽게 충족할 수 있습니다.
Self-RAG: Adaptive Retrieval and Self-Critique for LLMs
Self-RAG (ICLR 2024 Oral) trains a language model to decide when to retrieve and then grade its own results using four reflection tokens — reaching 55.8% on PopQA and 80.2 FactScore on biographies while outperforming ChatGPT on five benchmarks. Analysis covers the mechanism, ablation results, reproducibility limits, and implications for finance AI agents over Beancount ledgers.
Voyager: Skill Libraries as the Foundation for Lifelong AI Agent Learning
Voyager, a GPT-4-powered Minecraft agent from NVIDIA and Caltech, demonstrates that a persistent code skill library enables genuine lifelong learning without fine-tuning — discovering 3.3× more items than prior state-of-the-art. The pattern maps directly onto long-horizon Beancount ledger automation, though financial correctness demands staging layers that game sandboxes never require.
HippoRAG: Neurobiologically Inspired Long-Term Memory for LLMs
HippoRAG (NeurIPS 2024) builds a knowledge graph from OpenIE triples and applies Personalized PageRank at query time, reaching 89.1% Recall@5 on 2WikiMultiHopQA versus 68.2% for ColBERTv2—with direct implications for querying complex financial ledgers across multi-year transaction histories.