Internalizing Tool Knowledge in Small Language Models via QLoRA Fine-Tuning

• URL: http://arxiv.org/abs/2605.17774v2
• Date: Tue, 26 May 2026 01:46:08 GMT
• Title: Internalizing Tool Knowledge in Small Language Models via QLoRA Fine-Tuning
• Abstract summary: This paper investigates whether tool-use knowledge can be internalized into small language models.<n>We fine-tune Gemma 4 E4B and Qwen3-4B with 8-bit QLoRA on approximately 1,700 tool-use examples.<n>We evaluate the resulting models under description-free inference, where the prompt omits the tool catalog entirely.
• Score: 4.42060064555688
• License: http://creativecommons.org/publicdomain/zero/1.0/
• Abstract: Large language models are increasingly used as planning components in agentic systems, but current tool-use pipelines often require full tool schemas to be included in every prompt, creating substantial token overhead and limiting the practicality of smaller models. This paper investigates whether tool-use knowledge can be internalized into small language models through parameter-efficient fine-tuning, enabling structured planning without explicit tool descriptions at inference time. Using AssetOpsBench as the primary benchmark, we fine-tune Gemma 4 E4B and Qwen3-4B with 8-bit QLoRA on approximately 1,700 tool-use examples spanning tool knowledge, question-to-plan mappings, and execution-style traces. We evaluate the resulting models under description-free inference, where the prompt omits the tool catalog entirely. The fine-tuned models outperform an informed unfine-tuned baseline that receives full tool descriptions, reducing input length by 82.6\% while improving structural and LLM-judge planning scores. In the best Gemma run, the model achieves an AT-F1 of 0.65 and an overall judge score of 3.88, compared with 0.47 and 2.88 for the informed baseline. Qwen3-4B achieves a strong overall judge score of 3.78 while using 62\% less memory and running 2.5$\times$ faster than Gemma, though it also exhibits greater catastrophic forgetting on general multiple-choice benchmarks. Additional ablations show that LoRA rank controls a quality--retention trade-off, with $r=32$ maximizing planning quality and smaller ranks preserving more general knowledge. These results suggest that, for fixed tool catalogs, QLoRA fine-tuning can shift tool knowledge from prompt context into model weights, substantially reducing inference overhead while maintaining or improving tool-planning quality.

Related papers

• Joint Optimization of Tool Creation and Use for Large Language Model Agents [69.07066297088187]
  SMITH (-grounded Multi-task Iterative Tool Honing) trains tool creation and tool use inside a single policy.<n>A 4B Qwen3 trained with SMITH on 13 procedural reasoning tasks with exact verifiers reaches 79.8 macro-average accuracy on held-out tasks.<n>It also reaches 40.4 on TabMWP-Hard and 42.6 on out-of-domain GQA (+7.6 over the best same-backbone inference-time baseline)
  arXiv  Detail & Related papers  (2026-08-25T13:59:34Z)
• The Remarkable Effectiveness of Providing AI Agents with Natural Language Tools: A Replication Study Validating NLT Performance Across 14 Models [41.99844472131922]
  NLT improves tool-calling accuracy by 14.9 percentage points overall (62.3% versus 47.4% structured) and reduces critical errors by 93% (51 versus 755 errors)<n>NLT is a practical alternative to structured tool calling, especially for production systems that value reliability over parseability.
  arXiv  Detail & Related papers  (2026-07-04T16:59:12Z)
• SING: Synthetic Intention Graph for Scalable Active Tool Discovery in LLM Agents [64.59100414726556]
  Large language model (LLM) agents increasingly rely on harnesses that manage context, tools, and multi-turn execution.<n>Retrieval-augmented tool selection offers a natural alternative, but existing one-shot retrieval methods fail to align isolated tool descriptions with the agent's true task intention.<n>We propose SING, an intention-aware active tool discovery framework that builds an intention-tool graph linking user intentions, tool capabilities, and tool collaboration patterns.
  arXiv  Detail & Related papers  (2026-06-15T11:37:37Z)
• LLM Agents Already Know When to Call Tools -- Even Without Reasoning [25.40369702634587]
  LLM agents tend to call tools indiscriminately, even when the model can answer directly.<n>No existing benchmark systematically studies when a tool call is actually needed.<n>We propose When2Tool, a benchmark of 18 environments spanning three categories of tool necessity.
  arXiv  Detail & Related papers  (2026-05-10T01:37:40Z)
• Tool-Aware Planning in Contact Center AI: Evaluating LLMs through Lineage-Guided Query Decomposition [2.8180871881371456]
  We present a domain-grounded framework and benchmark for tool-aware plan generation in contact centers.<n>Our contributions are threefold: (i) a reference-based plan evaluation framework operating in two modes - a metric-wise evaluator and a one-shot evaluator, and (ii) a data methodology that iteratively refines plans via an evaluator->optimizer loop.
  arXiv  Detail & Related papers  (2026-02-16T17:36:05Z)
• MCP-Atlas: A Large-Scale Benchmark for Tool-Use Competency with Real MCP Servers [5.463884405989425]
  We introduce MCP-Atlas, a large-scale benchmark for evaluating tool-use competency.<n>It includes 1,000 tasks designed to assess tool-use competency in realistic, multi-step orchestrate.<n>We score tasks using a claims-based rubric that awards partial credit based on the factual claims satisfied in the model's final answer.
  arXiv  Detail & Related papers  (2026-01-31T23:19:39Z)
• One Model to Critique Them All: Rewarding Agentic Tool-Use via Efficient Reasoning [54.580646706013965]
  Reward models (RMs) play a critical role in aligning large language models with human preferences.<n>We introduce ToolRM, a family of lightweight generative RMs tailored for general tool-use scenarios.<n>To build these models, we propose a novel pipeline that constructs pairwise preference data using rule-based scoring and multidimensional sampling.
  arXiv  Detail & Related papers  (2025-10-30T06:08:27Z)
• OSWorld-MCP: Benchmarking MCP Tool Invocation In Computer-Use Agents [49.34040731113563]
  We present OSWorld-MCP, the first comprehensive and fair benchmark for assessing computer-use agents' tool invocation, GUI operation, and decision-making abilities.<n> Rigorous manual validation yields 158 high-quality tools, each verified for correct functionality, practical applicability, and versatility.<n> OSWorld-MCP deepens understanding of multimodal agents and sets a new standard for evaluating performance in complex, tool-assisted environments.
  arXiv  Detail & Related papers  (2025-10-28T15:56:36Z)
• Advancing Tool-Augmented Large Language Models via Meta-Verification and Reflection Learning [63.2198957755528]
  We propose Tool-MVR, a novel Tool-Augmented LLM that achieves comprehensive System 2 reasoning through two key innovations.<n>Specifically, we first introduce Multi-Agent Meta-Verification (MAMV), a systematic pipeline that rigorously validates APIs, queries, and reasoning trajectories.<n>Second, we propose Exploration-based Reflection Learning (EXPLORE), which enhances tool reflection capabilities by leveraging tool feedback.
  arXiv  Detail & Related papers  (2025-06-05T04:35:49Z)
• Acting Less is Reasoning More! Teaching Model to Act Efficiently [87.28134636548705]
  Tool-integrated reasoning augments large language models with the ability to invoke external tools to solve tasks.<n>Current approaches typically optimize only for final correctness without considering the efficiency or necessity of external tool use.<n>We propose a framework that encourages models to produce accurate answers with minimal tool calls.<n>Our approach reduces tool calls by up to 68.3% and improves tool productivity by up to 215.4%, while maintaining comparable answer accuracy.
  arXiv  Detail & Related papers  (2025-04-21T05:40:05Z)
• SOPBench: Evaluating Language Agents at Following Standard Operating Procedures and Constraints [59.645885492637845]
  SOPBench is an evaluation pipeline that transforms each service-specific SOP code program into a directed graph of executable functions.<n>Our approach transforms each service-specific SOP code program into a directed graph of executable functions and requires agents to call these functions based on natural language SOP descriptions.<n>We evaluate 18 leading models, and results show the task is challenging even for top-tier models.
  arXiv  Detail & Related papers  (2025-03-11T17:53:02Z)
• ToolBeHonest: A Multi-level Hallucination Diagnostic Benchmark for Tool-Augmented Large Language Models [43.895478182631116]
  Tool-augmented large language models (LLMs) are rapidly being integrated into real-world applications. To address this challenge, we introduce a comprehensive diagnostic benchmark, ToolBH. For breadth, we consider three scenarios based on the characteristics of the toolset: missing necessary tools, potential tools, and limited functionality tools. The results show the significant challenges presented by the ToolBH benchmark.
  arXiv  Detail & Related papers  (2024-06-28T16:03:30Z)
• ToolAlpaca: Generalized Tool Learning for Language Models with 3000 Simulated Cases [49.7798644853604]
  This paper introduces ToolAlpaca, a framework designed to automatically generate a diverse tool-use corpus and learn generalized tool-use abilities on compact language models. We show that ToolAlpaca achieves effective generalized tool-use capabilities comparable to those of extremely large language models like GPT-3.5.
  arXiv  Detail & Related papers  (2023-06-08T15:46:32Z)

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.