Tag
This paper introduces ToolSciVer, the first tool-augmented framework for multimodal scientific claim verification (MSCV), which equips a VLM with type-aware visual tools and trains the policy using GRPO to achieve superior performance on SciVer and MuSciClaims datasets across multiple model families.
This paper proposes UrbanAgent, an agentic framework that reframes urban region profiling as a reasoning-driven inference problem using multi-agent collaborative reasoning and tool-augmented evidence retrieval. It outperforms baselines on global urban datasets for carbon emissions, GDP, and population estimation, achieving an average 8.1% improvement in R².
This paper presents a taxonomy of LLM reasoning strategies along two orthogonal axes: fast vs. slow thinking and internal vs. external knowledge, and surveys recent adaptive reasoning methods.
This paper presents an empirical study and benchmark for evaluating tool-augmented LLM agents on real-world energy analytics tasks, comprising 243 expert-curated problems across market data retrieval, knowledge interpretation, and quantitative modeling.
This survey categorizes LLM-based optimization into three paradigms—direct, tool-augmented, and tool-creating—and reviews their performance frontiers and limitations.
This paper proposes Risk-Aware Causal Gating (RACG), a training-free mechanism that applies the principle of least privilege to LLM agent tool exposure, reducing attack surface from prompt injection by only exposing high-risk tools when authorized and causally necessary.
This paper introduces Contract2Tool, a framework for automatically inferring lightweight tool contracts (preconditions, effects, risk) from tool metadata, documentation, and execution traces, enabling reliable causal tool filtering for LLM agents. Experiments show learned contracts achieve near-gold contract performance in downstream multi-step agent tasks, significantly reducing token usage.
ToolGate is a lightweight external controller that predicts whether to execute or skip perceptual tool calls in vision-language agents, reducing token cost to 64–69% of baseline while preserving accuracy in cross-domain settings.
Introduces TADDLE, a tool-augmented agent for detecting deficient LLM-generated peer reviews, along with an expert-annotated benchmark of 1,800 reviews on 50 ICLR 2025 papers. The system decomposes detection into four specialized analysis tools and uses two-stage semi-supervised learning for binary and multi-label classification.
The paper introduces COSMO-Agent, a tool-augmented reinforcement learning framework that trains LLMs to perform closed-loop CAD-CAE optimization, iteratively generating parametric geometries and running simulations until constraints are satisfied, with a multi-constraint reward and a new industry-aligned dataset.
Introduces AutoTool, a model that adaptively decides whether to invoke tools for multimodal LLM reasoning, achieving significant accuracy and efficiency gains through reinforcement learning and dual-mode reasoning.
This paper proposes a lightweight, signal-based framework for efficiently triaging agentic interaction trajectories by computing low-cost indicators that identify informative samples without impacting online agent behavior, achieving an 82% informativeness rate on benchmarks.