arXiv:2511.12281v1 Announce Type: cross 
Abstract: Motivated by the high costs of using black-box Large Language Models (LLMs), we introduce a novel prompt compression paradigm, under which we use smaller LLMs to compress inputs for the larger ones. We present the first comprehensive LLM-as-a-compressor benchmark spanning 25 open- and closed-source models, which reveals significant disparity in models' compression ability in terms of (i) preserving semantically important information (ii) following the user-provided compression rate (CR). We further improve the performance of gpt-4.1-mini, the best overall vanilla compressor, with Textgrad-based compression meta-prompt optimization. We also identify the most promising open-source vanilla LLM - Qwen3-4B - and post-train it with a combination of supervised fine-tuning (SFT) and Group Relative Policy Optimization (GRPO), pursuing the dual objective of CR adherence and maximizing the downstream task performance. We call the resulting model Cmprsr and demonstrate its superiority over both extractive and vanilla abstractive compression across the entire range of compression rates on lengthy inputs from MeetingBank and LongBench as well as short prompts from GSM8k. The latter highlights Cmprsr's generalizability across varying input lengths and domains. Moreover, Cmprsr closely follows the requested compression rate, offering fine control over the cost-quality trade-off.

يقدم المقال Cmprsr، وهو نموذج جديد لضغط المطالبات يهدف إلى تقليل التكاليف المرتبطة باستخدام نماذج اللغة الكبيرة (LLMs) ذات الصندوق الأسود. من خلال استخدام نماذج لغة أصغر لضغط المدخلات للنماذج الأكبر، يقدم البحث معيارًا عبر 25 نموذجًا، مما يبرز الفروقات الكبيرة في قدرتها على الحفاظ على المعلومات المهمة والامتثال لمعدلات الضغط المحددة من قبل المستخدم. تم تحسين أداء gpt-4.1-mini من خلال تقنيات تحسين، بينما تم تحديد Qwen3-4B كنموذج مفتوح المصدر واعد تم تدريبه بعد ذلك لتحسين الأداء في المهام.

El artículo presenta Cmprsr, un nuevo paradigma de compresión de prompts que busca reducir los costos asociados con el uso de grandes modelos de lenguaje (LLM) de caja negra. Al utilizar modelos de lenguaje más pequeños para comprimir las entradas destinadas a modelos más grandes, el estudio presenta un banco de pruebas que abarca 25 modelos, destacando diferencias significativas en su capacidad para preservar información importante y cumplir con las tasas de compresión definidas por el usuario. Se mejora el rendimiento de gpt-4.1-mini mediante técnicas de optimización, mientras que Qwen3-4B s…

L'article présente Cmprsr, un nouveau paradigme de compression de prompts visant à réduire les coûts associés à l'utilisation de grands modèles de langage (LLM) en boîte noire. En utilisant de plus petits LLM pour compresser les entrées destinées à des modèles plus grands, l'étude présente une évaluation de 25 modèles, mettant en évidence des différences significatives dans leur capacité à préserver des informations importantes et à respecter les taux de compression définis par l'utilisateur. La performance de gpt-4.1-mini est améliorée grâce à des techniques d'optimisation, tandis que Qwen3-4…

The article introduces Cmprsr, a novel prompt compression paradigm aimed at reducing the costs associated with using large black-box language models (LLMs). By utilizing smaller LLMs to compress inputs for larger ones, the study presents a benchmark across 25 models, highlighting significant differences in their ability to preserve important information and adhere to user-defined compression rates. The performance of gpt-4.1-mini is enhanced through optimization techniques, while Qwen3-4B is identified as a promising open-source model that undergoes post-training to improve task performance.

Cmprsr: Abstractive Token-Level Question-Agnostic Prompt Compressor

arXiv:2601.08739v1 Announce Type: new 
Abstract: Knowledge graphs (KGs) provide structured evidence that can ground large language model (LLM) reasoning for knowledge-intensive question answering. However, many practical KGs are private, and sending retrieved triples or exploration traces to closed-source LLM APIs introduces leakage risk. Existing privacy treatments focus on masking entity names, but they still face four limitations: structural leakage under semantic masking, uncontrollable remote interaction, fragile multi-hop and multi-entity reasoning, and limited experience reuse for stability and efficiency. To address these issues, we propose PrivGemo, a privacy-preserving retrieval-augmented framework for KG-grounded reasoning with memory-guided exposure control. PrivGemo uses a dual-tower design to keep raw KG knowledge local while enabling remote reasoning over an anonymized view that goes beyond name masking to limit both semantic and structural exposure. PrivGemo supports multi-hop, multi-entity reasoning by retrieving anonymized long-hop paths that connect all topic entities, while keeping grounding and verification on the local KG. A hierarchical controller and a privacy-aware experience memory further reduce unnecessary exploration and remote interactions. Comprehensive experiments on six benchmarks show that PrivGemo achieves overall state-of-the-art results, outperforming the strongest baseline by up to 17.1%. Furthermore, PrivGemo enables smaller models (e.g., Qwen3-4B) to achieve reasoning performance comparable to that of GPT-4-Turbo.

تم تقديم PrivGemo كإطار يحافظ على الخصوصية مصممًا للمنطق القائم على الرسوم البيانية للمعرفة (KG)، حيث يعالج المخاطر المرتبطة باستخدام KGs الخاصة في نماذج اللغة الكبيرة (LLM). تحافظ هذه البنية ذات البرجين على المعرفة المحلية بينما تسمح بالتفكير عن بُعد من خلال واجهة مجهولة، مما يقلل بشكل فعال من التعرض الدلالي والهيكلي.

PrivGemo se ha presentado como un marco que preserva la privacidad, diseñado para el razonamiento basado en grafos de conocimiento (KG), abordando los riesgos asociados con el uso de KGs privados en modelos de lenguaje de gran tamaño (LLM). Esta arquitectura de doble torre mantiene el conocimiento local mientras permite el razonamiento remoto a través de una interfaz anonimizada, mitigando efectivamente la exposición semántica y estructural.

PrivGemo a été introduit comme un cadre préservant la vie privée, conçu pour le raisonnement basé sur des graphes de connaissances (KG), abordant les risques associés à l'utilisation de KG privés dans les modèles de langage de grande taille (LLM). Cette architecture à double tour maintient les connaissances locales tout en permettant un raisonnement à distance via une interface anonymisée, atténuant efficacement l'exposition sémantique et structurelle.

PrivGemo has been introduced as a privacy-preserving framework designed for knowledge graph (KG)-grounded reasoning, addressing the risks associated with using private KGs in large language models (LLMs). This dual-tower architecture maintains local knowledge while allowing remote reasoning through an anonymized interface, effectively mitigating semantic and structural exposure.

PrivGemo: Privacy-Preserving Dual-Tower Graph Retrieval for Empowering LLM Reasoning with Memory Augmentation

arXiv:2510.26167v2 Announce Type: replace-cross 
Abstract: Reward models (RMs) play a critical role in aligning large language models (LLMs) with human preferences. Yet in the domain of tool learning, the lack of RMs specifically designed for function-calling tasks has limited progress toward more capable agentic AI. We introduce ToolRM, a family of lightweight reward models tailored for general tool-use scenarios. To build these models, we propose a novel pipeline that constructs high-quality pairwise preference data using rule-based scoring and multidimensional sampling. This yields ToolPref-Pairwise-30K, a diverse, balanced, and challenging preference dataset that supports both generative and discriminative reward modeling. We also introduce TRBench$_{BFCL}$, a benchmark built on the agent evaluation suite BFCL to evaluate RMs on tool calling tasks. Trained on our constructed data, models from the Qwen3-4B/8B series achieve up to 17.94% higher accuracy, substantially outperforming frontier LLMs and RMs in pairwise reward judgments. Beyond training objectives, generative ToolRM generalizes to broader critique tasks, including Best-of-N sampling and self-correction. Experiments on ACEBench highlight its effectiveness and efficiency, enabling inference-time scaling while reducing output token usage by over 66%. Its support for downstream RL training further validates its practical utility. We release data to facilitate future research.

تم تقديم ToolRM كعائلة جديدة من نماذج المكافآت الخفيفة المصممة خصيصًا لسيناريوهات استخدام الأدوات، مما يعالج قيود نماذج المكافآت الحالية في مواءمة نماذج اللغة الكبيرة (LLMs) مع تفضيلات البشر. يتضمن هذا التطور خط أنابيب مبتكر لتوليد بيانات تفضيل عالية الجودة ومعيارًا لتقييم هذه النماذج في مهام استدعاء الأدوات.

ToolRM se ha presentado como una nueva familia de modelos de recompensa ligeros diseñados específicamente para escenarios de uso de herramientas, abordando las limitaciones de los modelos de recompensa existentes en la alineación de grandes modelos de lenguaje (LLMs) con las preferencias humanas. Este desarrollo incluye un nuevo pipeline para generar datos de preferencias de alta calidad y un benchmark para evaluar estos modelos en tareas de llamada de herramientas.

ToolRM a été introduit comme une nouvelle famille de modèles de récompense légers spécifiquement conçus pour les scénarios d'utilisation d'outils, répondant aux limitations des modèles de récompense existants dans l'alignement des grands modèles de langage (LLMs) avec les préférences humaines. Ce développement comprend un pipeline novateur pour générer des données de préférence de haute qualité et un benchmark pour évaluer ces modèles sur des tâches d'appel d'outils.

ToolRM has been introduced as a new family of lightweight reward models specifically designed for tool-use scenarios, addressing the limitations of existing reward models in aligning large language models (LLMs) with human preferences. This development includes a novel pipeline for generating high-quality preference data and a benchmark for evaluating these models on tool-calling tasks.

Cmprsr: Abstractive Token-Level Question-Agnostic Prompt Compressor

Was this article worth reading? Share it

LucidQuery AI

Airparser

AQ

Sellm

Supametas.AI

Langfuse

Ready to build your own newsroom?