# Model Fusion with Multi-LoRA Inference for Tool-Enhanced Game Dialogue Agents

arXiv: https://arxiv.org/abs/2509.24229 · HTML: https://mlsanity.com/paper/2509.24229/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 7 of its 9 references that are in the map.
- First posted on arXiv 29 Sep 2025.

## Builds on (most cited first, 7 of 7)

- [BLEURT: Learning Robust Metrics for Text Generation](https://mlsanity.com/paper/2004.04696/) (2020)
- [SWIFT:A Scalable lightWeight Infrastructure for Fine-Tuning](https://mlsanity.com/paper/2408.05517/) (2024)
- [SOLAR 10.7B: Scaling Large Language Models with Simple yet Effective Depth Up-Scaling](https://mlsanity.com/paper/2312.15166/) (2023)
- [xLAM: A Family of Large Action Models to Empower AI Agent Systems](https://mlsanity.com/paper/2409.03215/) (2024)
- [PeaCoK: Persona Commonsense Knowledge for Consistent and Engaging Narratives](https://mlsanity.com/paper/2305.02364/) (2023)
- [ComperDial: Commonsense Persona-grounded Dialogue Dataset and Benchmark](https://mlsanity.com/paper/2406.11228/) (2024)
- [Winning Amazon KDD Cup'24](https://mlsanity.com/paper/2408.04658/) (2024)

## Built on by (most cited first, 1 of 1)

- [PReCache: Efficient KV Cache Sharing for Multi-LoRA Agents via Low-Rank Precomputation and Neutral Reconstruction](https://mlsanity.com/paper/2609.34054/) (2026)

## Abstract (arXiv)

This paper presents the opdainlp team's solution for the GPU track of the CPDC 2025 challenge. The challenge consists of three tasks, aiming to build an in-game conversational AI that adheres to character personas, aligns with the game's worldview, and supports function calling. Considering both effectiveness and resource/time constraints during inference, we synthesized data for some of the tasks based on the datasets provided by the competition organizers. We employed Qwen3-14B with LoRA fine-tuning and model fusion, and utilized a base model integrated with multiple LoRA adapters during inference. Specifically, in the competition, we used three distinct LoRA adapters to handle tool calling, response generation with tool call results, and response generation without tool call results, respectively. MultiLoRA inference was implemented using vLLM. Our solution achieved the first place in Task 1 and Task 3, and the second place in Task 2 of the GPU track.