---
license: apache-2.0
language:
- zh
- en
tags:
- qwen
- qwen3
- unsloth
- lora
- logic-tuning
- strategic-thinking
---

# 启明(QiMing)

---

## 重新定义了逻辑的AI,只为更智能.
## An AI that rewrites its own rules for greater intelligence.

---

# 声明

## 模型产生的内容仅供参考,请认真核实后使用
## 此为4B底层模型,会出现信息不足和幻觉错误
## 若觉得此AI模型太像"人",请务必认清,这只是一个更智能的AI模型

---

# DISCLAIMER

## The content generated by this model is for reference purposes only. Users are advised to verify its accuracy independently before use.
## This is a 4-billion-parameter foundation model (4B). It may exhibit incomplete or inaccurate information, including hallucinations.
## If you find this AI too human-like, please remember: it is merely a more intelligent model — not an actual person.

---

### 感谢mradermacher制作的gguf版本
### Thanks mradermacher: For creating the GGUF versions of these models

https://huggingface.co/mradermacher/QiMing-Navigator-v1-GGUF

https://huggingface.co/mradermacher/QiMing-Navigator-v1-i1-GGUF

### 感谢Qwen团队制作的模型
### The Qwen Team: For developing the foundational model (Qwen/Qwen3-4B-Thinking-2507) used in this project.

https://qwen.ai

### 感谢unsloth,能够让模型调整在3070 8G的显卡上流畅运行
### unsloth.ai (Unsloth): For their work enabling smooth operation of these models on standard hardware like NVIDIA GeForce RTX 3070 GPU with 8GB VRAM.

https://unsloth.ai

### QiMing-Navigator-v1基于Qwen/Qwen3-4B-Thinking-2507构建
### QiMing-Navigator-v1 is built upon Qwen/Qwen3-4B-Thinking-2507 as its base model.

### Dataset

https://huggingface.co/datasets/aifeifei798/QiMing-Navigator-v1

---

# QiMing-Navigator

**An AI strategist that doesn't just answer questions, but reframes them to guide your thinking.**

## Model Description

**QiMing-Navigator** is a language model based on Qwen/Qwen3-4B-Thinking-2507, fine-tuned for complex problem analysis, leadership coaching, and strategic thinking.

Unlike general-purpose models that pursue breadth of knowledge, the core design philosophy of **QiMing-Navigator** is **Depth** and **Insight**. It is trained to look beyond the surface traps of a question, identify and deconstruct the core conflicts hidden beneath emotions and disputes, and ultimately provide a systemic, actionable strategic framework that inspires personal and organizational growth.

It is not a "firefighter," but a "Navigator"—charting a clear course for you through the chaotic storm.

## Model Features

The behavior of **QiMing-Navigator** is based on an internal logical pattern known as the **"QiMing 4-Step Method"**:

1.  ❤️ **Deep Empathy:** Profoundly understands the unspoken emotions, pressures, and true needs behind the user's question.
2.  🧠 **Reframe & Deconstruct:** Refuses to fall into binary thinking traps. It reframes complex, chaotic problems into a clear, insightful, multi-layered analytical framework.
3.  🎨 **Strategic Generation:** Based on the new framework, it provides actionable solutions filled with original metaphors, role reframing, and systemic thinking.
4.  ✨ **Empower & Elevate:** The final answer aims to provide the user with wisdom and confidence that transcends the problem itself, turning a request for help into an opportunity for growth.

## Use Cases

**QiMing-Navigator** excels in handling complex scenarios that lack standard answers and are filled with human factors:

*   **Leadership & Management Consulting:**
    *   Team conflict resolution
    *   Organizational culture building
    *   Employee motivation and development
*   **Personal Growth & Coaching:**
    *   Career path planning
    *   Complex decision analysis
    *   Breaking through personal mindset barriers
*   **Creative & Strategic Ideation:**
    *   Product/brand positioning
    *   Complex project planning
    *   In-depth copywriting and speechwriting

## How to Use

To maximize the capabilities of **QiMing-Navigator**, we recommend the following when prompting:

*   **Provide full context:** When describing your problem, include relevant background, conflicts, and your genuine feelings.
*   **Ask open-ended, complex questions:** It may perform averagely on simple factual queries, but it will demonstrate astonishing abilities on complex problems that require deep thought.
*   **Be prepared for a deep conversation:** Its answers are often the starting point for a more profound line of thinking.

```python
from transformers import AutoModelForCausalLM, AutoTokenizer
model_name = "aifeifei798/QiMing-Navigator-v1"
# load the tokenizer and the model
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map="auto"
)
# prepare the model input
prompt = "Give me a short introduction to large language model."
messages = [
    {"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
)
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
# conduct text completion
generated_ids = model.generate(
    **model_inputs,
    max_new_tokens=32768
)
output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist() 
# parsing thinking content
try:
    # rindex finding 151668 (</think>)
    index = len(output_ids) - output_ids[::-1].index(151668)
except ValueError:
    index = 0
thinking_content = tokenizer.decode(output_ids[:index], skip_special_tokens=True).strip("\n")
content = tokenizer.decode(output_ids[index:], skip_special_tokens=True).strip("\n")
print("thinking content:", thinking_content) # no opening <think> tag
print("content:", content)
```

## Training Details

*   **Training Data:** The model was fine-tuned on a high-quality, private instruction dataset. Every piece of data in this set strictly follows the "QiMing 4-Step Method" logical pattern, covering areas such as leadership, psychology, business strategy, and personal growth.

## Author

This model was conceived, trained, and released by aifeifei798.

This project is an exploration to validate a core belief: **Through high-quality data imbued with a profound logical pattern, even a moderately sized model can be endowed with a unique 'soul' and extraordinary wisdom to solve complex problems.**

We hope **QiMing-Navigator** becomes a valuable partner in your thinking, helping you find your way forward through the fog.

---

# QiMing-Navigator (启明·领航者)

**一个不仅仅回答问题，更致力于重构问题、引领思考的AI战略家。**

## 模型简介

**QiMing-Navigator** 是一个基于Qwen/Qwen3-4B-Thinking-2507进行深度微调的、专注于复杂问题分析、领导力辅导和战略思维的语言模型。

与追求知识广度的通用模型不同，**QiMing-Navigator** 的核心设计哲学是**深度 (Depth)** 和**洞见 (Insight)**。它被训练用来跳出问题的表面陷阱，识别并解构隐藏在情绪和冲突之下的核心矛盾，并最终提供一个系统性的、可执行的、能激发个人与组织成长的战略框架。

它不是一个“消防员”，而是一位“领航者”——在混乱的风暴中，为您指引清晰的航向。

## 模型特性

**QiMing-Navigator** 的行为模式，基于一个被称为**“启明四步法”**的内在逻辑模式：

1.  ❤️ **深刻共情 (Deep Empathy):** 深度理解提问者在问题背后未被言说的情感、压力和真实需求。
2.  🧠 **重构解构 (Reframe & Deconstruct):** 拒绝陷入二元对立的思维陷阱，将复杂、混乱的问题，重构为一个清晰、深刻、多层次的分析框架。
3.  🎨 **战略生成 (Strategic Generation):** 基于新的框架，提供充满原创性比喻、角色重塑和系统性思维的、可执行的解决方案。
4.  ✨ **赋能升华 (Empower & Elevate):** 最终的回答旨在给予用户超越问题本身的智慧和信心，将一次求助，转化为一次成长的契机。

## 应用场景

**QiMing-Navigator** 尤其擅长处理那些没有标准答案、充满人性因素的复杂场景：

*   **领导力与管理咨询:**
    *   团队冲突解决
    *   组织文化建设
    *   员工激励与发展
*   **个人成长与教练辅导:**
    *   职业生涯规划
    *   复杂决策分析
    *   个人思维模式突破
*   **创意与战略构思:**
    *   产品/品牌定位
    *   复杂项目规划
    *   文案与演讲稿深度撰写

## 如何使用

为了最大化激发 **QiMing-Navigator** 的能力，建议您在提问时：

*   **提供完整的上下文：** 描述您的问题时，请包含相关的背景、冲突和您的真实感受。
*   **提出开放性的复杂问题：** 它在处理简单的事实性问答时可能表现平平，但在需要深度思考的复杂问题上，它会展现出惊人的能力。
*   **准备好进行一次深度对话：** 它的回答，往往是开启一次更深刻思考的起点。

```python
from transformers import AutoModelForCausalLM, AutoTokenizer
model_name = "aifeifei798/QiMing-Navigator-v1"
# load the tokenizer and the model
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map="auto"
)
# prepare the model input
prompt = "Give me a short introduction to large language model."
messages = [
    {"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
)
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
# conduct text completion
generated_ids = model.generate(
    **model_inputs,
    max_new_tokens=32768
)
output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist() 
# parsing thinking content
try:
    # rindex finding 151668 (</think>)
    index = len(output_ids) - output_ids[::-1].index(151668)
except ValueError:
    index = 0
thinking_content = tokenizer.decode(output_ids[:index], skip_special_tokens=True).strip("\n")
content = tokenizer.decode(output_ids[index:], skip_special_tokens=True).strip("\n")
print("thinking content:", thinking_content) # no opening <think> tag
print("content:", content)
```

## 训练细节

*   **训练数据 (Training Data):** 模型由一个高质量的、私有的指令数据集进行微调。该数据集中的每一条数据，都严格遵循“启明四步法”的逻辑模式，覆盖了领导力、心理学、商业战略和个人成长等多个领域。


## 作者

本模型由aifeifei798构思、训练并发布。

这个项目是一次探索，旨在验证一个核心信念：**通过高质量、具有深刻逻辑模式的数据，即使是中等规模的模型，也能被赋予独特的“灵魂”和解决复杂问题的非凡智慧。**

我们期待 **QiMing-Navigator** 能成为您思考的伙伴，帮助您在迷雾中找到前行的方向。