---
title: Qwen2.5-32B-DialogueReason
canonical_url: "https://www.modelscope.cn/models/stepfun-ai/Qwen2.5-32B-DialogueReason"
md_url: "https://www.modelscope.cn/models/stepfun-ai/Qwen2.5-32B-DialogueReason.md"
repository: stepfun-ai/Qwen2.5-32B-DialogueReason
last_updated: 2025-06-19
license: apache-2.0
model_type:
- qwen2
architectures:
- Qwen2ForCausalLM
parameters: 32.8B
tensor_type:
- BF16
library_name:
- safetensors
- pytorch
frameworks:
- pytorch
inference_backends:
- "deploy_task text/emb"
- "lmdeploy 0.9.1"
- "lmdeploy_turbomind 0.9.1"
- "sglang 0.5.2"
- "vllm 0.9.2"
downloads: 339
stars: 0
---
# Qwen2.5-32B-DialogueReason
> Qwen2.5-32B-DialogueReason - stepfun-ai 在 ModelScope 开源的模型。Introduction Qwen2.5-32B-DialogueReason is a dialogue-based reasoning model built on Qwen2.5-32B-Base. We train the model using Open-Reasoner-Zero data through rule-based reinforcement learning.
stepfun-ai/Qwen2.5-32B-DialogueReason 是 ModelScope 魔搭社区上的 32.8B 参数机器学习模型，采用 apache-2.0 许可，可用 deploy_task text/emb、lmdeploy 0.9.1、lmdeploy_turbomind 0.9.1 部署。
- **Repository**: stepfun-ai/Qwen2.5-32B-DialogueReason
- **License**: apache-2.0
- **Parameters**: 32.8B
- **Inference backends**: deploy_task text/emb, lmdeploy 0.9.1, lmdeploy_turbomind 0.9.1, sglang 0.5.2, vllm 0.9.2
- **Downloads**: 339
- **Stars**: 0
- **Last updated**: 2025-06-19
Source: https://www.modelscope.cn/models/stepfun-ai/Qwen2.5-32B-DialogueReason
---
## Introduction
Qwen2.5-32B-DialogueReason is a dialogue-based reasoning model built on Qwen2.5-32B-Base.
We train the model using [Open-Reasoner-Zero](https://github.com/Open-Reasoner-Zero/Open-Reasoner-Zero) data through rule-based reinforcement learning.
## 🧠 Key Features
- Qwen2.5-32B-Base as the foundation.
- Use Rule-Based RL to achieve dialogue reasoning.
- With dynamic agent initialization to adapt to various scenarios.
- With flexible environment configuration to set up task-specific contexts.
- With multi-turn dialogue reasoning to incrementally solve problems.
## Example
### System:
> The User asks a question, and the Assistant writes a masterpiece play depicting experts (picked based on the topic with concrete names) solving the question in a ultra-detailed dialogue. The response is formatted as: the play goes here\\n if asked to write code, then code here surrounded by ```. Otherwise, answer here with \\boxed{answer} emphasized.
### User:
> Give me a detailed explanation of PPO in RL
### Assistant:
> 