FrankenReport: Early Exiting in Long-Form Generation Using Expected Value of Computation

• URL: http://arxiv.org/abs/2609.06320v1
• Date: Sun, 06 Sep 2026 00:43:52 GMT
• Title: FrankenReport: Early Exiting in Long-Form Generation Using Expected Value of Computation
• Abstract summary: We present FrankenReport, an interface for long-form knowledge-seeking report generation that supports adaptive early exiting per section.<n>In a simulation study, FrankenReport outperforms random allocation baselines by a large margin (up to 4x) under low budgets.<n>We show that despite varying preferences across users and topics, FrankenReport adapts to simple, natural user feedback as efficiently as methods requiring much costlier supervision.
• Score: 57.70956612390847
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: While deep research systems address interactive information-seeking needs impressively, their real-world deployments face latency and resource-consumption challenges. We present FrankenReport, an interface for long-form knowledge-seeking report generation that supports adaptive early exiting per section: it evaluates intermediate outputs during generation and predicts whether further targeted computation will yield significant quality gains. In a simulation study, FrankenReport outperforms random allocation baselines by a large margin (up to 4x) under low budgets and smoothly recovers full-pipeline quality as the budget grows, showing that future quality gains are predictable from intermediate drafts. Through experiments and user studies, we further show that despite varying preferences across users and topics, FrankenReport adapts to simple, natural user feedback as efficiently as methods requiring much costlier supervision such as generated drafts and explicit rationales.

Related papers

• Unleashing the Potential of Large Language Models: A Blueprint for Real-Time, Enterprise-Ready Deployments [11.71713024399078]
  We present a unified, pattern-driven LLMOps architecture integrating real-time data ingestion, continual learning, retrieval-augmented generation, and human-in-the-loop feedback.<n>The result reduces latency-cost-accuracy trade-offs while supporting auditability and rollback for high-risk sectors such as health care and finance.
  arXiv  Detail & Related papers  (2026-08-01T03:40:22Z)
• Tabular Foundation Models for Discrete Choice Estimation [8.167439804974467]
  Tabular foundation models (TFMs) generate predictions on structured data via in-context learning, without task-specific estimation.<n>We ask whether TFMs can be effectively applied to discrete choice, a central demand estimation framework in marketing and operations.<n>We propose a reformulation that encodes both choice-set dependence and individual heterogeneity within a row-based learning framework.
  arXiv  Detail & Related papers  (2026-07-14T22:34:00Z)
• TRACE: A Unified Rollout Budget Allocation Framework for Efficient Agentic Reinforcement Learning [52.67756371231985]
  Reinforcement learning with verifiable rewards (RLVR) is a promising approach for enhancing reasoning and agentic behavior in large language models.<n>We introduce Tree Rollout Allocation for Contrastive Exploration (TRACE), a unified rollout allocation framework.<n>Technically, TRACE allocates rollout budget to both prompt roots and intermediate prefixes that are most likely to yield mixed terminal rewards.
  arXiv  Detail & Related papers  (2026-06-09T17:16:03Z)
• Data-dependent Exploration for Online Reinforcement Learning from Human Feedback [50.34161049551627]
  Online reinforcement learning from human feedback (RLHF) has emerged as a promising paradigm for aligning large language models (LLMs) by continuously collecting new preference feedback during training.<n>Existing exploration strategies often derive bonuses via on-policy expectations, which are difficult to estimate reliably from the limited historical preference data available during training.<n>We propose data-dependent exploration for preference optimization (DEPO), a simple and scalable method that leverages historical data to construct an extra uncertainty bonus for high-uncertainty regions.
  arXiv  Detail & Related papers  (2026-05-06T03:56:45Z)
• LiveFMBench: Unveiling the Power and Limits of Agentic Workflows in Specification Generation [75.05397479715576]
  Large language models (LLMs) and agents have shown promising progress, but their true capabilities and failure modes remain unclear.<n>We present the first systematic and contamination-aware study of LLM- and agent-based formal specification generation for C programs.
  arXiv  Detail & Related papers  (2026-05-02T11:31:33Z)
• Efficient Dataset Selection for Continual Adaptation of Generative Recommenders [19.8068072624233]
  This work investigates how targeted data selection can mitigate performance degradation caused by temporal distributional drift.<n>We evaluate a range of representation choices and sampling strategies for curating small but informative subsets of user interaction data.
  arXiv  Detail & Related papers  (2026-04-09T02:48:52Z)
• Learning User Interests via Reasoning and Distillation for Cross-Domain News Recommendation [7.070021001906444]
  News recommendation plays a critical role in online news platforms by helping users discover relevant content.<n>Cross-domain news recommendation further requires inferring user's underlying information needs from heterogeneous signals.<n>We present a reinforcement learning framework that trains large language models to generate high-quality lists of interest-driven news search queries.
  arXiv  Detail & Related papers  (2026-02-16T18:45:40Z)
• SMA: Who Said That? Auditing Membership Leakage in Semi-Black-box RAG Controlling [50.66950115630554]
  Retrieval-Augmented Generation (RAG) and its Multimodal Retrieval-Augmented Generation (MRAG) significantly improve the knowledge coverage and contextual understanding of Large Language Models (LLMs)<n>However, retrieval and multimodal fusion obscure content provenance, rendering existing membership inference methods unable to reliably attribute generated outputs to pre-training, external retrieval, or user input, thus undermining privacy leakage accountability.<n>We propose the first Source-aware Membership Audit (SMA) that enables fine-grained source attribution of generated content in a semi-black-box setting with retrieval control capabilities.
  arXiv  Detail & Related papers  (2025-08-12T17:32:24Z)
• Temporal Window Smoothing of Exogenous Variables for Improved Time Series Prediction [6.300598034097709]
  We propose a method that whitens the endogenous input to reduce redundancy that may persist within the data based on global statistics.<n>Our approach achieves state-of-the-art performance in four benchmark datasets, consistently outperforming 11 baseline models.
  arXiv  Detail & Related papers  (2025-07-04T17:27:55Z)
• ZigZag: Universal Sampling-free Uncertainty Estimation Through Two-Step Inference [54.17205151960878]
  We introduce a sampling-free approach that is generic and easy to deploy. We produce reliable uncertainty estimates on par with state-of-the-art methods at a significantly lower computational cost.
  arXiv  Detail & Related papers  (2022-11-21T13:23:09Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.