|
|
--- |
|
|
license: mit |
|
|
pipeline_tag: text-generation |
|
|
library_name: transformers |
|
|
--- |
|
|
|
|
|
--- |
|
|
frameworks: |
|
|
- Pytorch |
|
|
license: mit |
|
|
tasks: |
|
|
- text-generation |
|
|
language: |
|
|
- en |
|
|
metrics: |
|
|
- accuracy |
|
|
base_model: |
|
|
- Qwen/Qwen2.5-0.5B-Instruct |
|
|
|
|
|
|
|
|
# Model Card for Tool-Star |
|
|
|
|
|
This is the official checkpoint we trained using the tool-star framework, based on Qwen2.5-0.5B-Instruct. |
|
|
|
|
|
Huggingface Paper: https://huggingface.co/papers/2505.16410 |
|
|
|
|
|
Details please refer to https://github.com/dongguanting/Tool-Star |
|
|
|
|
|
# Paper title and link |
|
|
|
|
|
The model was presented in the paper [Tool-Star: Empowering LLM-Brained Multi-Tool Reasoner via Reinforcement |
|
|
Learning](https://huggingface.co/papers/2505.16410). |
|
|
|
|
|
# Paper abstract |
|
|
|
|
|
The abstract of the paper is the following: |
|
|
|
|
|
Recently, large language models (LLMs) have shown remarkable reasoning |
|
|
capabilities via large-scale reinforcement learning (RL). However, leveraging |
|
|
the RL algorithm to empower effective multi-tool collaborative reasoning in |
|
|
LLMs remains an open challenge. In this paper, we introduce Tool-Star, an |
|
|
RL-based framework designed to empower LLMs to autonomously invoke multiple |
|
|
external tools during stepwise reasoning. Tool-Star integrates six types of |
|
|
tools and incorporates systematic designs in both data synthesis and training. |
|
|
To address the scarcity of tool-use data, we propose a general tool-integrated |
|
|
reasoning data synthesis pipeline, which combines tool-integrated prompting |
|
|
with hint-based sampling to automatically and scalably generate tool-use |
|
|
trajectories. A subsequent quality normalization and difficulty-aware |
|
|
classification process filters out low-quality samples and organizes the |
|
|
dataset from easy to hard. Furthermore, we propose a two-stage training |
|
|
framework to enhance multi-tool collaborative reasoning by: (1) cold-start |
|
|
fine-tuning, which guides LLMs to explore reasoning patterns via |
|
|
tool-invocation feedback; and (2) a multi-tool self-critic RL algorithm with |
|
|
hierarchical reward design, which reinforces reward understanding and promotes |
|
|
effective tool collaboration. Experimental analyses on over 10 challenging |
|
|
reasoning benchmarks highlight the effectiveness and efficiency of Tool-Star. |
|
|
The code is available at https://github.com/dongguanting/Tool-Star. |
