|
|
--- |
|
|
base_model: |
|
|
- Qwen/Qwen3-4B-Instruct-2507 |
|
|
language: |
|
|
- zh |
|
|
- en |
|
|
license: apache-2.0 |
|
|
metrics: |
|
|
- accuracy |
|
|
- f1 |
|
|
pipeline_tag: text-generation |
|
|
library_name: transformers |
|
|
--- |
|
|
|
|
|
# TCAndon-Router |
|
|
|
|
|
<p align="center"> |
|
|
<img src="https://raw.githubusercontent.com/Tencent/TCAndon-Router/refs/heads/main/assets/router.png" width="500"/> |
|
|
</p> |
|
|
|
|
|
<p align="center"> |
|
|
<a href="https://github.com/Tencent/TCAndon-Router">Github</a> | 📑 <a href="https://arxiv.org/pdf/2601.04544">Paper</a> |
|
|
</p> |
|
|
|
|
|
|
|
|
## 🌟 Introduction |
|
|
|
|
|
In multi-agent systems, the ability to select the appropriate agent(s) to handle a user query is a key determinant of overall system performance. |
|
|
|
|
|
TCAndonRouter is a reasoning-centric multi-intent routing module whose primary role is to perform agent routing in multi-agent systems. |
|
|
Beyond agent routing, TCAndonRouter can be applied to any intent-routing scenario, including agent skill selection. |
|
|
|
|
|
The main advantages of TCAndonRouter include: |
|
|
|
|
|
+ Designed specifically for real-world enterprise applications |
|
|
+ Supports dynamic onboarding of new agents (intents) New agents can be added simply by appending their descriptions, without retraining |
|
|
+ Provides transparent and interpretable routing decisions, improving explainability, robustness, and cross-domain generalization, and making post-deployment bad-case analysis easier |
|
|
+ Effectively resolves agent conflicts caused by overlapping responsibilities, leading to higher-quality final responses. When multiple agents are applicable, TCAndonRouter preserves all relevant agents. Each downstream agent generates its own response, and a Refining Agent subsequently merges these outputs into a single final answer |
|
|
|
|
|
TCAndonRouter is trained using Supervised Fine-Tuning (SFT) followed by Reinforcement Learning (DAPO), and achieves state-of-the-art performance on large-scale, real-world enterprise datasets, including HWU64, MINDS14, SGD, and the Tencent Cloud ITSM dataset(QCloud). |
|
|
|
|
|
| **Models** | **CLINC150** | **HWU64** | **MINDS14** | **SGD** | **QCloud** | |
|
|
|------------------------|--------------|-----------|-------------|-----------|-----------------| |
|
|
| GPT-5.1 | 93.84 | 85.59 | 95.59 | 73.90 | 92.80/93.06 | |
|
|
| Claude-Sonnet-4.5 | **94.21** | 87.40 | 96.20 | 76.02 | 88.82/94.25 | |
|
|
| DeepSeek-v3.1-terminus | 88.29 | 88.10 | 95.72 | 79.70 | 94.09/91.89 | |
|
|
| ArcRouter | 62.98 | 69.33 | 91.79 | 65.59 | - | |
|
|
| Qwen3-Embedding-4B | 57.21 | 54.27 | 94.12 | 37.02 | - | |
|
|
| Qwen3-4B-Instruct-2507 | 70.12 | 80.29 | 90.08 | 58.74 | 82.23/79.44 | |
|
|
| **TCAndonRouter** | 91.25 | **91.63** | **96.70** | **91.58** | **95.21/92.78** | |
|
|
|
|
|
## 🔧 How to use |
|
|
|
|
|
Please refer to [GitHub](https://github.com/Tencent/TCAndon-Router) for code usage. |
|
|
|
|
|
```python |
|
|
from transformers import AutoModelForCausalLM, AutoTokenizer |
|
|
from prompt import router_prompt |
|
|
from utils import load_config |
|
|
|
|
|
tokenizer = AutoTokenizer.from_pretrained("tencent/TCAndon-Router") |
|
|
model = AutoModelForCausalLM.from_pretrained("tencent/TCAndon-Router", device_map="auto") |
|
|
|
|
|
agents = load_config('config/hwu64_config.xml') |
|
|
query = "Can you recommend any pub in mg road" |
|
|
prompt = router_prompt.format(agents=agents) + 'user:' + query |
|
|
|
|
|
messages = [{"role": "user", "content": prompt}] |
|
|
encoding = tokenizer.apply_chat_template( |
|
|
messages, |
|
|
tokenize=True, |
|
|
add_generation_prompt=False, |
|
|
return_tensors="pt" |
|
|
) |
|
|
|
|
|
outputs = model.generate(encoding.to(model.device), max_new_tokens=2048) |
|
|
output_text = tokenizer.decode(outputs[0]) |
|
|
``` |
|
|
### Generate Agent Descriptions |
|
|
|
|
|
If you want to use TCAndonRouter on your own dataset, you need to provide agent descriptions. The required format is defined in `config/xxx_config.xml`. |
|
|
You can generate agent descriptions using an LLM via generate_agent_desc.py, or write them manually. |
|
|
```shell |
|
|
python generate_agent_desc.py --dataset hwu64 --limit 50 |
|
|
``` |
|
|
|
|
|
## 🤝 Citation |
|
|
|
|
|
If you use TCAndonRouter in your work, please cite our paper: |
|
|
|
|
|
``` |
|
|
@article{zhao2026TCAndonRouter, |
|
|
title={TCAndonRouter: Adaptive Reasoning Router for Multi-Agent Collaboration}, |
|
|
author={Jiuzhou Zhao, Chunrong Chen, Chenqi Qiao, Lebin Zheng, Minqi Han, Yanchi Liu, Yongzhou Xu, Xiaochuan Xu, Min Zhang}, |
|
|
journal={arXiv preprint:2601.04544}, |
|
|
year={2026} |
|
|
} |
|
|
``` |
