README.md

---
language:
- en
license: apache-2.0
size_categories:
- 100K<n<1M
tags:
- function-calling
- LLM Agent
- reward-modeling
---

<h1 align="center">ToolRM Training Dataset</h1>

<div align="center">
<a width="150" style="display: inline-block" href="https://arxiv.org/abs/2509.11963"><img alt="Static Badge" src="https://img.shields.io/badge/arxiv-2509.11963-red?logo=arxiv"></a>
<a width="150" style="display: inline-block" href="https://huggingface.co/datasets/ibm-research/fc-reward-bench"><img alt="Static Badge" src="https://img.shields.io/badge/HF-fc--reward--bench-green?logo=huggingface"></a>
</div>

## 📖 Dataset Description

This is a version of the training data utilized for ToolRM, a collection of outcome reward models specifically designed for evaluating and improving function-calling capabilities in large language models. It consists of ~459K examples, where each example includes a user-assistant conversation, available tool specifications, and a pair of correct and incorrect tool calls. The incorrect calls were generated by prompting 9 open-source language models on queries from three public datasets. Reward Models trained on this dataset were found to result in an average improvement of up to 25% in downstream task performance, enhance robustness to input noise, and enable data-efficient fine-tuning through reward-guided filtering.



## 📊 Dataset Statistics

- **Total Training Samples**: 458,575
- **Composition**:
  - Single-turn interactions: 256,851 samples
  - Multi-turn interactions: 159,757 samples  
  - Irrelevance cases: 41,967 samples
- **Source Datasets**: [APIGen](https://huggingface.co/datasets/Salesforce/xlam-function-calling-60k), [Schema-Guided Dialogue (SGD)](https://github.com/google-research-datasets/dstc8-schema-guided-dialogue), [xlam-irrelevance](https://huggingface.co/datasets/MadeAgents/xlam-irrelevance-7.5k)
- **Generator Models**: 9 permissively-licensed open-weight models


## 🗂️ Dataset Schema

The dataset contains the following fields:

| Field | Type | Description |
|-------|------|-------------|
| `uuid` | str | Unique identifier for each training sample |
| `dataset_name` | str | Source dataset from which the sample was derived |
| `conversation` | list | Conversation between user and assistant |
| `tools` | str | Catalog of available function specifications |
| `tool_calls_correct` | str | Ground-truth correct tool invocations for the given conversation |
| `tool_calls_incorrect` | str | Incorrect tool invocations generated by the model pool |
| `generator_model` | str | Identifier of the model that produced the incorrect tool call |

*Note: `tools`, `tool_calls_correct`, and `tool_calls_incorrect` fields have been serialized. While loading the dataset, convert them into JSON objects using `json.loads`*


## ⚙️ Data Generation Methodology

### Generator Model Pool

The incorrect tool calls were generated using the following models:

- **Granite Series**: [granite-3.3-2b-instruct](https://huggingface.co/ibm-granite/granite-3.2-2b-instruct), [granite-3.3-8b-instruct](https://huggingface.co/ibm-granite/granite-3.2-8b-instruct), [granite-20b-functioncalling](https://huggingface.co/ibm-granite/granite-20b-functioncalling)
- **SmolLM**: [SmolLM2-1.7B-Instruct](https://huggingface.co/HuggingFaceTB/SmolLM2-1.7B-Instruct), [SmolLM3-3B](https://huggingface.co/HuggingFaceTB/SmolLM3-3B)
- **Mistral Series**: [Mistral-7B-Instruct-v0.3](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.3), [Mistral-Nemo-Instruct-2407](https://huggingface.co/mistralai/Mistral-Nemo-Instruct-2407)
- **GPT-OSS Series**: [gpt-oss-20b](https://huggingface.co/openai/gpt-oss-20b), [gpt-oss-120b](https://huggingface.co/openai/gpt-oss-120b)

### Data Collection Process

1. **Source Datasets**: We start with publicly available function-calling datasets that cover a wide range of interaction patterns
2. **Obfuscation**: Function and parameter names were replaced with random strings, and schema keys were reordered to prevent models from regurgitating the training data
3. **Generation**: Each sample is processed through the model pool to generate function calls
4. **Verification**: The generated outputs are compared against ground-truth annotations to identify incorrect calls
5. **Filtering**: We keep only the incorrect generations, selecting up to three incorrect samples per query to maintain diversity while avoiding over-representation


## 🎯 Benchmark

In a Best-of-N setting, we found that ToolRM significantly improves performance over Greedy decoding, Majority Voting, and Schema Validation baselines.

<div align="center">
<img src="https://cdn-uploads.huggingface.co/production/uploads/6229237ed94a4a3d5efbacb5/m-I-B9TSRKq-CtpuQWW5C.png" width=800 />
</div>

For reward-guided data filtering, we found that a model fine-tuned with 8K top-ranked samples by ToolRM outperforms the model fine-tuned with the entire training dataset of 16K samples.

<div align="center">
<img src="https://cdn-uploads.huggingface.co/production/uploads/6229237ed94a4a3d5efbacb5/Dq3_-yPlvOFxQTjf_Mi2a.png" width=800 />
</div>

More experiments and a detailed discussion of the results can be found in the paper.

## 📚 Citation

If you use this dataset in your research, please cite:

```
@misc{agarwal2025toolrmoutcomereward,
      title={ToolRM: Outcome Reward Models for Tool-Calling Large Language Models},
      author={Mayank Agarwal and Ibrahim Abdelaziz and Kinjal Basu and Merve Unuvar and Luis A. Lastras and Yara Rizk and Pavan Kapanipathi},
      year={2025},
      eprint={2509.11963},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2509.11963},
}
```